fix math

2.1.782

.github/CODEOWNERS

@@ -0,0 +1 @@
+*       @lllyasviel

.github/ISSUE_TEMPLATE/bug_report.md

@@ -0,0 +1,14 @@
+---
+name: Bug report
+about: Describe a problem
+title: ''
+labels: ''
+assignees: ''
+
+---
+
+**Describe the problem**
+A clear and concise description of what the bug is.
+
+**Full Console Log**
+Paste **full** console log here. You will make our job easier if you give a **full** log.

.github/ISSUE_TEMPLATE/feature_request.md

@@ -0,0 +1,14 @@
+---
+name: Feature request
+about: Suggest an idea for this project
+title: ''
+labels: ''
+assignees: ''
+
+---
+
+**Is your feature request related to a problem? Please describe.**
+A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]
+
+**Describe the idea you'd like**
+A clear and concise description of what you want to happen.

.gitignore

@@ -2,7 +2,22 @@ __pycache__
 *.ckpt
 *.safetensors
 *.pth
-!taesdxl_decoder.pth
+*.pt
+*.bin
+*.patch
+*.backup
+*.corrupted
+/language/default.json
+lena.png
+lena_result.png
+lena_test.py
+config.txt
+config_modification_tutorial.txt
+user_path_config.txt
+user_path_config-deprecated.txt
+build_chb.py
+experiment.py
+/modules/*.png
 /repositories
 /venv
 /tmp
@@ -20,7 +35,8 @@ __pycache__
 /interrogate
 /user.css
 /.idea
-notification.mp3
+/notification.ogg
+/notification.mp3
 /SwinIR
 /textual_inversion
 .vscode
@@ -32,3 +48,4 @@ notification.mp3
 /node_modules
 /package-lock.json
 /.coverage*
+/auth.json

LICENSE

@@ -0,0 +1,674 @@
+                    GNU GENERAL PUBLIC LICENSE
+                       Version 3, 29 June 2007
+
+ Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+                            Preamble
+
+  The GNU General Public License is a free, copyleft license for
+software and other kinds of works.
+
+  The licenses for most software and other practical works are designed
+to take away your freedom to share and change the works.  By contrast,
+the GNU General Public License is intended to guarantee your freedom to
+share and change all versions of a program--to make sure it remains free
+software for all its users.  We, the Free Software Foundation, use the
+GNU General Public License for most of our software; it applies also to
+any other work released this way by its authors.  You can apply it to
+your programs, too.
+
+  When we speak of free software, we are referring to freedom, not
+price.  Our General Public Licenses are designed to make sure that you
+have the freedom to distribute copies of free software (and charge for
+them if you wish), that you receive source code or can get it if you
+want it, that you can change the software or use pieces of it in new
+free programs, and that you know you can do these things.
+
+  To protect your rights, we need to prevent others from denying you
+these rights or asking you to surrender the rights.  Therefore, you have
+certain responsibilities if you distribute copies of the software, or if
+you modify it: responsibilities to respect the freedom of others.
+
+  For example, if you distribute copies of such a program, whether
+gratis or for a fee, you must pass on to the recipients the same
+freedoms that you received.  You must make sure that they, too, receive
+or can get the source code.  And you must show them these terms so they
+know their rights.
+
+  Developers that use the GNU GPL protect your rights with two steps:
+(1) assert copyright on the software, and (2) offer you this License
+giving you legal permission to copy, distribute and/or modify it.
+
+  For the developers' and authors' protection, the GPL clearly explains
+that there is no warranty for this free software.  For both users' and
+authors' sake, the GPL requires that modified versions be marked as
+changed, so that their problems will not be attributed erroneously to
+authors of previous versions.
+
+  Some devices are designed to deny users access to install or run
+modified versions of the software inside them, although the manufacturer
+can do so.  This is fundamentally incompatible with the aim of
+protecting users' freedom to change the software.  The systematic
+pattern of such abuse occurs in the area of products for individuals to
+use, which is precisely where it is most unacceptable.  Therefore, we
+have designed this version of the GPL to prohibit the practice for those
+products.  If such problems arise substantially in other domains, we
+stand ready to extend this provision to those domains in future versions
+of the GPL, as needed to protect the freedom of users.
+
+  Finally, every program is threatened constantly by software patents.
+States should not allow patents to restrict development and use of
+software on general-purpose computers, but in those that do, we wish to
+avoid the special danger that patents applied to a free program could
+make it effectively proprietary.  To prevent this, the GPL assures that
+patents cannot be used to render the program non-free.
+
+  The precise terms and conditions for copying, distribution and
+modification follow.
+
+                       TERMS AND CONDITIONS
+
+  0. Definitions.
+
+  "This License" refers to version 3 of the GNU General Public License.
+
+  "Copyright" also means copyright-like laws that apply to other kinds of
+works, such as semiconductor masks.
+
+  "The Program" refers to any copyrightable work licensed under this
+License.  Each licensee is addressed as "you".  "Licensees" and
+"recipients" may be individuals or organizations.
+
+  To "modify" a work means to copy from or adapt all or part of the work
+in a fashion requiring copyright permission, other than the making of an
+exact copy.  The resulting work is called a "modified version" of the
+earlier work or a work "based on" the earlier work.
+
+  A "covered work" means either the unmodified Program or a work based
+on the Program.
+
+  To "propagate" a work means to do anything with it that, without
+permission, would make you directly or secondarily liable for
+infringement under applicable copyright law, except executing it on a
+computer or modifying a private copy.  Propagation includes copying,
+distribution (with or without modification), making available to the
+public, and in some countries other activities as well.
+
+  To "convey" a work means any kind of propagation that enables other
+parties to make or receive copies.  Mere interaction with a user through
+a computer network, with no transfer of a copy, is not conveying.
+
+  An interactive user interface displays "Appropriate Legal Notices"
+to the extent that it includes a convenient and prominently visible
+feature that (1) displays an appropriate copyright notice, and (2)
+tells the user that there is no warranty for the work (except to the
+extent that warranties are provided), that licensees may convey the
+work under this License, and how to view a copy of this License.  If
+the interface presents a list of user commands or options, such as a
+menu, a prominent item in the list meets this criterion.
+
+  1. Source Code.
+
+  The "source code" for a work means the preferred form of the work
+for making modifications to it.  "Object code" means any non-source
+form of a work.
+
+  A "Standard Interface" means an interface that either is an official
+standard defined by a recognized standards body, or, in the case of
+interfaces specified for a particular programming language, one that
+is widely used among developers working in that language.
+
+  The "System Libraries" of an executable work include anything, other
+than the work as a whole, that (a) is included in the normal form of
+packaging a Major Component, but which is not part of that Major
+Component, and (b) serves only to enable use of the work with that
+Major Component, or to implement a Standard Interface for which an
+implementation is available to the public in source code form.  A
+"Major Component", in this context, means a major essential component
+(kernel, window system, and so on) of the specific operating system
+(if any) on which the executable work runs, or a compiler used to
+produce the work, or an object code interpreter used to run it.
+
+  The "Corresponding Source" for a work in object code form means all
+the source code needed to generate, install, and (for an executable
+work) run the object code and to modify the work, including scripts to
+control those activities.  However, it does not include the work's
+System Libraries, or general-purpose tools or generally available free
+programs which are used unmodified in performing those activities but
+which are not part of the work.  For example, Corresponding Source
+includes interface definition files associated with source files for
+the work, and the source code for shared libraries and dynamically
+linked subprograms that the work is specifically designed to require,
+such as by intimate data communication or control flow between those
+subprograms and other parts of the work.
+
+  The Corresponding Source need not include anything that users
+can regenerate automatically from other parts of the Corresponding
+Source.
+
+  The Corresponding Source for a work in source code form is that
+same work.
+
+  2. Basic Permissions.
+
+  All rights granted under this License are granted for the term of
+copyright on the Program, and are irrevocable provided the stated
+conditions are met.  This License explicitly affirms your unlimited
+permission to run the unmodified Program.  The output from running a
+covered work is covered by this License only if the output, given its
+content, constitutes a covered work.  This License acknowledges your
+rights of fair use or other equivalent, as provided by copyright law.
+
+  You may make, run and propagate covered works that you do not
+convey, without conditions so long as your license otherwise remains
+in force.  You may convey covered works to others for the sole purpose
+of having them make modifications exclusively for you, or provide you
+with facilities for running those works, provided that you comply with
+the terms of this License in conveying all material for which you do
+not control copyright.  Those thus making or running the covered works
+for you must do so exclusively on your behalf, under your direction
+and control, on terms that prohibit them from making any copies of
+your copyrighted material outside their relationship with you.
+
+  Conveying under any other circumstances is permitted solely under
+the conditions stated below.  Sublicensing is not allowed; section 10
+makes it unnecessary.
+
+  3. Protecting Users' Legal Rights From Anti-Circumvention Law.
+
+  No covered work shall be deemed part of an effective technological
+measure under any applicable law fulfilling obligations under article
+11 of the WIPO copyright treaty adopted on 20 December 1996, or
+similar laws prohibiting or restricting circumvention of such
+measures.
+
+  When you convey a covered work, you waive any legal power to forbid
+circumvention of technological measures to the extent such circumvention
+is effected by exercising rights under this License with respect to
+the covered work, and you disclaim any intention to limit operation or
+modification of the work as a means of enforcing, against the work's
+users, your or third parties' legal rights to forbid circumvention of
+technological measures.
+
+  4. Conveying Verbatim Copies.
+
+  You may convey verbatim copies of the Program's source code as you
+receive it, in any medium, provided that you conspicuously and
+appropriately publish on each copy an appropriate copyright notice;
+keep intact all notices stating that this License and any
+non-permissive terms added in accord with section 7 apply to the code;
+keep intact all notices of the absence of any warranty; and give all
+recipients a copy of this License along with the Program.
+
+  You may charge any price or no price for each copy that you convey,
+and you may offer support or warranty protection for a fee.
+
+  5. Conveying Modified Source Versions.
+
+  You may convey a work based on the Program, or the modifications to
+produce it from the Program, in the form of source code under the
+terms of section 4, provided that you also meet all of these conditions:
+
+    a) The work must carry prominent notices stating that you modified
+    it, and giving a relevant date.
+
+    b) The work must carry prominent notices stating that it is
+    released under this License and any conditions added under section
+    7.  This requirement modifies the requirement in section 4 to
+    "keep intact all notices".
+
+    c) You must license the entire work, as a whole, under this
+    License to anyone who comes into possession of a copy.  This
+    License will therefore apply, along with any applicable section 7
+    additional terms, to the whole of the work, and all its parts,
+    regardless of how they are packaged.  This License gives no
+    permission to license the work in any other way, but it does not
+    invalidate such permission if you have separately received it.
+
+    d) If the work has interactive user interfaces, each must display
+    Appropriate Legal Notices; however, if the Program has interactive
+    interfaces that do not display Appropriate Legal Notices, your
+    work need not make them do so.
+
+  A compilation of a covered work with other separate and independent
+works, which are not by their nature extensions of the covered work,
+and which are not combined with it such as to form a larger program,
+in or on a volume of a storage or distribution medium, is called an
+"aggregate" if the compilation and its resulting copyright are not
+used to limit the access or legal rights of the compilation's users
+beyond what the individual works permit.  Inclusion of a covered work
+in an aggregate does not cause this License to apply to the other
+parts of the aggregate.
+
+  6. Conveying Non-Source Forms.
+
+  You may convey a covered work in object code form under the terms
+of sections 4 and 5, provided that you also convey the
+machine-readable Corresponding Source under the terms of this License,
+in one of these ways:
+
+    a) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by the
+    Corresponding Source fixed on a durable physical medium
+    customarily used for software interchange.
+
+    b) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by a
+    written offer, valid for at least three years and valid for as
+    long as you offer spare parts or customer support for that product
+    model, to give anyone who possesses the object code either (1) a
+    copy of the Corresponding Source for all the software in the
+    product that is covered by this License, on a durable physical
+    medium customarily used for software interchange, for a price no
+    more than your reasonable cost of physically performing this
+    conveying of source, or (2) access to copy the
+    Corresponding Source from a network server at no charge.
+
+    c) Convey individual copies of the object code with a copy of the
+    written offer to provide the Corresponding Source.  This
+    alternative is allowed only occasionally and noncommercially, and
+    only if you received the object code with such an offer, in accord
+    with subsection 6b.
+
+    d) Convey the object code by offering access from a designated
+    place (gratis or for a charge), and offer equivalent access to the
+    Corresponding Source in the same way through the same place at no
+    further charge.  You need not require recipients to copy the
+    Corresponding Source along with the object code.  If the place to
+    copy the object code is a network server, the Corresponding Source
+    may be on a different server (operated by you or a third party)
+    that supports equivalent copying facilities, provided you maintain
+    clear directions next to the object code saying where to find the
+    Corresponding Source.  Regardless of what server hosts the
+    Corresponding Source, you remain obligated to ensure that it is
+    available for as long as needed to satisfy these requirements.
+
+    e) Convey the object code using peer-to-peer transmission, provided
+    you inform other peers where the object code and Corresponding
+    Source of the work are being offered to the general public at no
+    charge under subsection 6d.
+
+  A separable portion of the object code, whose source code is excluded
+from the Corresponding Source as a System Library, need not be
+included in conveying the object code work.
+
+  A "User Product" is either (1) a "consumer product", which means any
+tangible personal property which is normally used for personal, family,
+or household purposes, or (2) anything designed or sold for incorporation
+into a dwelling.  In determining whether a product is a consumer product,
+doubtful cases shall be resolved in favor of coverage.  For a particular
+product received by a particular user, "normally used" refers to a
+typical or common use of that class of product, regardless of the status
+of the particular user or of the way in which the particular user
+actually uses, or expects or is expected to use, the product.  A product
+is a consumer product regardless of whether the product has substantial
+commercial, industrial or non-consumer uses, unless such uses represent
+the only significant mode of use of the product.
+
+  "Installation Information" for a User Product means any methods,
+procedures, authorization keys, or other information required to install
+and execute modified versions of a covered work in that User Product from
+a modified version of its Corresponding Source.  The information must
+suffice to ensure that the continued functioning of the modified object
+code is in no case prevented or interfered with solely because
+modification has been made.
+
+  If you convey an object code work under this section in, or with, or
+specifically for use in, a User Product, and the conveying occurs as
+part of a transaction in which the right of possession and use of the
+User Product is transferred to the recipient in perpetuity or for a
+fixed term (regardless of how the transaction is characterized), the
+Corresponding Source conveyed under this section must be accompanied
+by the Installation Information.  But this requirement does not apply
+if neither you nor any third party retains the ability to install
+modified object code on the User Product (for example, the work has
+been installed in ROM).
+
+  The requirement to provide Installation Information does not include a
+requirement to continue to provide support service, warranty, or updates
+for a work that has been modified or installed by the recipient, or for
+the User Product in which it has been modified or installed.  Access to a
+network may be denied when the modification itself materially and
+adversely affects the operation of the network or violates the rules and
+protocols for communication across the network.
+
+  Corresponding Source conveyed, and Installation Information provided,
+in accord with this section must be in a format that is publicly
+documented (and with an implementation available to the public in
+source code form), and must require no special password or key for
+unpacking, reading or copying.
+
+  7. Additional Terms.
+
+  "Additional permissions" are terms that supplement the terms of this
+License by making exceptions from one or more of its conditions.
+Additional permissions that are applicable to the entire Program shall
+be treated as though they were included in this License, to the extent
+that they are valid under applicable law.  If additional permissions
+apply only to part of the Program, that part may be used separately
+under those permissions, but the entire Program remains governed by
+this License without regard to the additional permissions.
+
+  When you convey a copy of a covered work, you may at your option
+remove any additional permissions from that copy, or from any part of
+it.  (Additional permissions may be written to require their own
+removal in certain cases when you modify the work.)  You may place
+additional permissions on material, added by you to a covered work,
+for which you have or can give appropriate copyright permission.
+
+  Notwithstanding any other provision of this License, for material you
+add to a covered work, you may (if authorized by the copyright holders of
+that material) supplement the terms of this License with terms:
+
+    a) Disclaiming warranty or limiting liability differently from the
+    terms of sections 15 and 16 of this License; or
+
+    b) Requiring preservation of specified reasonable legal notices or
+    author attributions in that material or in the Appropriate Legal
+    Notices displayed by works containing it; or
+
+    c) Prohibiting misrepresentation of the origin of that material, or
+    requiring that modified versions of such material be marked in
+    reasonable ways as different from the original version; or
+
+    d) Limiting the use for publicity purposes of names of licensors or
+    authors of the material; or
+
+    e) Declining to grant rights under trademark law for use of some
+    trade names, trademarks, or service marks; or
+
+    f) Requiring indemnification of licensors and authors of that
+    material by anyone who conveys the material (or modified versions of
+    it) with contractual assumptions of liability to the recipient, for
+    any liability that these contractual assumptions directly impose on
+    those licensors and authors.
+
+  All other non-permissive additional terms are considered "further
+restrictions" within the meaning of section 10.  If the Program as you
+received it, or any part of it, contains a notice stating that it is
+governed by this License along with a term that is a further
+restriction, you may remove that term.  If a license document contains
+a further restriction but permits relicensing or conveying under this
+License, you may add to a covered work material governed by the terms
+of that license document, provided that the further restriction does
+not survive such relicensing or conveying.
+
+  If you add terms to a covered work in accord with this section, you
+must place, in the relevant source files, a statement of the
+additional terms that apply to those files, or a notice indicating
+where to find the applicable terms.
+
+  Additional terms, permissive or non-permissive, may be stated in the
+form of a separately written license, or stated as exceptions;
+the above requirements apply either way.
+
+  8. Termination.
+
+  You may not propagate or modify a covered work except as expressly
+provided under this License.  Any attempt otherwise to propagate or
+modify it is void, and will automatically terminate your rights under
+this License (including any patent licenses granted under the third
+paragraph of section 11).
+
+  However, if you cease all violation of this License, then your
+license from a particular copyright holder is reinstated (a)
+provisionally, unless and until the copyright holder explicitly and
+finally terminates your license, and (b) permanently, if the copyright
+holder fails to notify you of the violation by some reasonable means
+prior to 60 days after the cessation.
+
+  Moreover, your license from a particular copyright holder is
+reinstated permanently if the copyright holder notifies you of the
+violation by some reasonable means, this is the first time you have
+received notice of violation of this License (for any work) from that
+copyright holder, and you cure the violation prior to 30 days after
+your receipt of the notice.
+
+  Termination of your rights under this section does not terminate the
+licenses of parties who have received copies or rights from you under
+this License.  If your rights have been terminated and not permanently
+reinstated, you do not qualify to receive new licenses for the same
+material under section 10.
+
+  9. Acceptance Not Required for Having Copies.
+
+  You are not required to accept this License in order to receive or
+run a copy of the Program.  Ancillary propagation of a covered work
+occurring solely as a consequence of using peer-to-peer transmission
+to receive a copy likewise does not require acceptance.  However,
+nothing other than this License grants you permission to propagate or
+modify any covered work.  These actions infringe copyright if you do
+not accept this License.  Therefore, by modifying or propagating a
+covered work, you indicate your acceptance of this License to do so.
+
+  10. Automatic Licensing of Downstream Recipients.
+
+  Each time you convey a covered work, the recipient automatically
+receives a license from the original licensors, to run, modify and
+propagate that work, subject to this License.  You are not responsible
+for enforcing compliance by third parties with this License.
+
+  An "entity transaction" is a transaction transferring control of an
+organization, or substantially all assets of one, or subdividing an
+organization, or merging organizations.  If propagation of a covered
+work results from an entity transaction, each party to that
+transaction who receives a copy of the work also receives whatever
+licenses to the work the party's predecessor in interest had or could
+give under the previous paragraph, plus a right to possession of the
+Corresponding Source of the work from the predecessor in interest, if
+the predecessor has it or can get it with reasonable efforts.
+
+  You may not impose any further restrictions on the exercise of the
+rights granted or affirmed under this License.  For example, you may
+not impose a license fee, royalty, or other charge for exercise of
+rights granted under this License, and you may not initiate litigation
+(including a cross-claim or counterclaim in a lawsuit) alleging that
+any patent claim is infringed by making, using, selling, offering for
+sale, or importing the Program or any portion of it.
+
+  11. Patents.
+
+  A "contributor" is a copyright holder who authorizes use under this
+License of the Program or a work on which the Program is based.  The
+work thus licensed is called the contributor's "contributor version".
+
+  A contributor's "essential patent claims" are all patent claims
+owned or controlled by the contributor, whether already acquired or
+hereafter acquired, that would be infringed by some manner, permitted
+by this License, of making, using, or selling its contributor version,
+but do not include claims that would be infringed only as a
+consequence of further modification of the contributor version.  For
+purposes of this definition, "control" includes the right to grant
+patent sublicenses in a manner consistent with the requirements of
+this License.
+
+  Each contributor grants you a non-exclusive, worldwide, royalty-free
+patent license under the contributor's essential patent claims, to
+make, use, sell, offer for sale, import and otherwise run, modify and
+propagate the contents of its contributor version.
+
+  In the following three paragraphs, a "patent license" is any express
+agreement or commitment, however denominated, not to enforce a patent
+(such as an express permission to practice a patent or covenant not to
+sue for patent infringement).  To "grant" such a patent license to a
+party means to make such an agreement or commitment not to enforce a
+patent against the party.
+
+  If you convey a covered work, knowingly relying on a patent license,
+and the Corresponding Source of the work is not available for anyone
+to copy, free of charge and under the terms of this License, through a
+publicly available network server or other readily accessible means,
+then you must either (1) cause the Corresponding Source to be so
+available, or (2) arrange to deprive yourself of the benefit of the
+patent license for this particular work, or (3) arrange, in a manner
+consistent with the requirements of this License, to extend the patent
+license to downstream recipients.  "Knowingly relying" means you have
+actual knowledge that, but for the patent license, your conveying the
+covered work in a country, or your recipient's use of the covered work
+in a country, would infringe one or more identifiable patents in that
+country that you have reason to believe are valid.
+
+  If, pursuant to or in connection with a single transaction or
+arrangement, you convey, or propagate by procuring conveyance of, a
+covered work, and grant a patent license to some of the parties
+receiving the covered work authorizing them to use, propagate, modify
+or convey a specific copy of the covered work, then the patent license
+you grant is automatically extended to all recipients of the covered
+work and works based on it.
+
+  A patent license is "discriminatory" if it does not include within
+the scope of its coverage, prohibits the exercise of, or is
+conditioned on the non-exercise of one or more of the rights that are
+specifically granted under this License.  You may not convey a covered
+work if you are a party to an arrangement with a third party that is
+in the business of distributing software, under which you make payment
+to the third party based on the extent of your activity of conveying
+the work, and under which the third party grants, to any of the
+parties who would receive the covered work from you, a discriminatory
+patent license (a) in connection with copies of the covered work
+conveyed by you (or copies made from those copies), or (b) primarily
+for and in connection with specific products or compilations that
+contain the covered work, unless you entered into that arrangement,
+or that patent license was granted, prior to 28 March 2007.
+
+  Nothing in this License shall be construed as excluding or limiting
+any implied license or other defenses to infringement that may
+otherwise be available to you under applicable patent law.
+
+  12. No Surrender of Others' Freedom.
+
+  If conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License.  If you cannot convey a
+covered work so as to satisfy simultaneously your obligations under this
+License and any other pertinent obligations, then as a consequence you may
+not convey it at all.  For example, if you agree to terms that obligate you
+to collect a royalty for further conveying from those to whom you convey
+the Program, the only way you could satisfy both those terms and this
+License would be to refrain entirely from conveying the Program.
+
+  13. Use with the GNU Affero General Public License.
+
+  Notwithstanding any other provision of this License, you have
+permission to link or combine any covered work with a work licensed
+under version 3 of the GNU Affero General Public License into a single
+combined work, and to convey the resulting work.  The terms of this
+License will continue to apply to the part which is the covered work,
+but the special requirements of the GNU Affero General Public License,
+section 13, concerning interaction through a network will apply to the
+combination as such.
+
+  14. Revised Versions of this License.
+
+  The Free Software Foundation may publish revised and/or new versions of
+the GNU General Public License from time to time.  Such new versions will
+be similar in spirit to the present version, but may differ in detail to
+address new problems or concerns.
+
+  Each version is given a distinguishing version number.  If the
+Program specifies that a certain numbered version of the GNU General
+Public License "or any later version" applies to it, you have the
+option of following the terms and conditions either of that numbered
+version or of any later version published by the Free Software
+Foundation.  If the Program does not specify a version number of the
+GNU General Public License, you may choose any version ever published
+by the Free Software Foundation.
+
+  If the Program specifies that a proxy can decide which future
+versions of the GNU General Public License can be used, that proxy's
+public statement of acceptance of a version permanently authorizes you
+to choose that version for the Program.
+
+  Later license versions may give you additional or different
+permissions.  However, no additional obligations are imposed on any
+author or copyright holder as a result of your choosing to follow a
+later version.
+
+  15. Disclaimer of Warranty.
+
+  THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
+APPLICABLE LAW.  EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
+HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
+OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
+THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+PURPOSE.  THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
+IS WITH YOU.  SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
+ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
+
+  16. Limitation of Liability.
+
+  IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
+THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
+GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
+USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
+DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
+PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
+EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
+SUCH DAMAGES.
+
+  17. Interpretation of Sections 15 and 16.
+
+  If the disclaimer of warranty and limitation of liability provided
+above cannot be given local legal effect according to their terms,
+reviewing courts shall apply local law that most closely approximates
+an absolute waiver of all civil liability in connection with the
+Program, unless a warranty or assumption of liability accompanies a
+copy of the Program in return for a fee.
+
+                     END OF TERMS AND CONDITIONS
+
+            How to Apply These Terms to Your New Programs
+
+  If you develop a new program, and you want it to be of the greatest
+possible use to the public, the best way to achieve this is to make it
+free software which everyone can redistribute and change under these terms.
+
+  To do so, attach the following notices to the program.  It is safest
+to attach them to the start of each source file to most effectively
+state the exclusion of warranty; and each file should have at least
+the "copyright" line and a pointer to where the full notice is found.
+
+    <one line to give the program's name and a brief idea of what it does.>
+    Copyright (C) <year>  <name of author>
+
+    This program is free software: you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+Also add information on how to contact you by electronic and paper mail.
+
+  If the program does terminal interaction, make it output a short
+notice like this when it starts in an interactive mode:
+
+    <program>  Copyright (C) <year>  <name of author>
+    This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
+    This is free software, and you are welcome to redistribute it
+    under certain conditions; type `show c' for details.
+
+The hypothetical commands `show w' and `show c' should show the appropriate
+parts of the General Public License.  Of course, your program's commands
+might be different; for a GUI interface, you would use an "about box".
+
+  You should also get your employer (if you work as a programmer) or school,
+if any, to sign a "copyright disclaimer" for the program, if necessary.
+For more information on this, and how to apply and follow the GNU GPL, see
+<https://www.gnu.org/licenses/>.
+
+  The GNU General Public License does not permit incorporating your program
+into proprietary programs.  If your program is a subroutine library, you
+may consider it more useful to permit linking proprietary applications with
+the library.  If this is what you want to do, use the GNU Lesser General
+Public License instead of this License.  But first, please read
+<https://www.gnu.org/licenses/why-not-lgpl.html>.

args_manager.py

@@ -0,0 +1,31 @@
+from fcbh.options import enable_args_parsing
+enable_args_parsing(False)
+import fcbh.cli_args as fcbh_cli
+
+
+fcbh_cli.parser.add_argument("--share", action='store_true', help="Set whether to share on Gradio.")
+fcbh_cli.parser.add_argument("--preset", type=str, default=None, help="Apply specified UI preset.")
+
+fcbh_cli.parser.add_argument("--language", type=str, default='default',
+                             help="Translate UI using json files in [language] folder. "
+                                  "For example, [--language example] will use [language/example.json] for translation.")
+
+# For example, https://github.com/lllyasviel/Fooocus/issues/849
+fcbh_cli.parser.add_argument("--enable-smart-memory", action="store_true",
+                             help="Force loading models to vram when the unload can be avoided. "
+                                  "Some Mac users may need this.")
+
+fcbh_cli.parser.set_defaults(
+    disable_cuda_malloc=True,
+    auto_launch=True,
+    port=None
+)
+
+fcbh_cli.args = fcbh_cli.parser.parse_args()
+
+# (beta, enabled by default. )
+# (Probably disable by default because of issues like https://github.com/lllyasviel/Fooocus/issues/724)
+if fcbh_cli.args.enable_smart_memory:
+    fcbh_cli.args.disable_smart_memory = False
+
+args = fcbh_cli.args

auth-example.json

@@ -0,0 +1,6 @@
+[
+    {
+        "user": "sitting-duck-1",
+        "pass": "very-bad-publicly-known-password-change-it"
+    }
+]

backend/doc

@@ -0,0 +1,5 @@
+# Fooocus' Comfy Backend Headless (FCBH)
+
+This is a Comfy Backend from StabilityAI. This pre-complied backend makes it easier for people who have trouble using pygit2.
+
+FCBH is maintained by Fooocus's reviewing upon StabilityAI's changes.

backend/headless/LICENSE

@@ -0,0 +1,674 @@
+                    GNU GENERAL PUBLIC LICENSE
+                       Version 3, 29 June 2007
+
+ Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+                            Preamble
+
+  The GNU General Public License is a free, copyleft license for
+software and other kinds of works.
+
+  The licenses for most software and other practical works are designed
+to take away your freedom to share and change the works.  By contrast,
+the GNU General Public License is intended to guarantee your freedom to
+share and change all versions of a program--to make sure it remains free
+software for all its users.  We, the Free Software Foundation, use the
+GNU General Public License for most of our software; it applies also to
+any other work released this way by its authors.  You can apply it to
+your programs, too.
+
+  When we speak of free software, we are referring to freedom, not
+price.  Our General Public Licenses are designed to make sure that you
+have the freedom to distribute copies of free software (and charge for
+them if you wish), that you receive source code or can get it if you
+want it, that you can change the software or use pieces of it in new
+free programs, and that you know you can do these things.
+
+  To protect your rights, we need to prevent others from denying you
+these rights or asking you to surrender the rights.  Therefore, you have
+certain responsibilities if you distribute copies of the software, or if
+you modify it: responsibilities to respect the freedom of others.
+
+  For example, if you distribute copies of such a program, whether
+gratis or for a fee, you must pass on to the recipients the same
+freedoms that you received.  You must make sure that they, too, receive
+or can get the source code.  And you must show them these terms so they
+know their rights.
+
+  Developers that use the GNU GPL protect your rights with two steps:
+(1) assert copyright on the software, and (2) offer you this License
+giving you legal permission to copy, distribute and/or modify it.
+
+  For the developers' and authors' protection, the GPL clearly explains
+that there is no warranty for this free software.  For both users' and
+authors' sake, the GPL requires that modified versions be marked as
+changed, so that their problems will not be attributed erroneously to
+authors of previous versions.
+
+  Some devices are designed to deny users access to install or run
+modified versions of the software inside them, although the manufacturer
+can do so.  This is fundamentally incompatible with the aim of
+protecting users' freedom to change the software.  The systematic
+pattern of such abuse occurs in the area of products for individuals to
+use, which is precisely where it is most unacceptable.  Therefore, we
+have designed this version of the GPL to prohibit the practice for those
+products.  If such problems arise substantially in other domains, we
+stand ready to extend this provision to those domains in future versions
+of the GPL, as needed to protect the freedom of users.
+
+  Finally, every program is threatened constantly by software patents.
+States should not allow patents to restrict development and use of
+software on general-purpose computers, but in those that do, we wish to
+avoid the special danger that patents applied to a free program could
+make it effectively proprietary.  To prevent this, the GPL assures that
+patents cannot be used to render the program non-free.
+
+  The precise terms and conditions for copying, distribution and
+modification follow.
+
+                       TERMS AND CONDITIONS
+
+  0. Definitions.
+
+  "This License" refers to version 3 of the GNU General Public License.
+
+  "Copyright" also means copyright-like laws that apply to other kinds of
+works, such as semiconductor masks.
+
+  "The Program" refers to any copyrightable work licensed under this
+License.  Each licensee is addressed as "you".  "Licensees" and
+"recipients" may be individuals or organizations.
+
+  To "modify" a work means to copy from or adapt all or part of the work
+in a fashion requiring copyright permission, other than the making of an
+exact copy.  The resulting work is called a "modified version" of the
+earlier work or a work "based on" the earlier work.
+
+  A "covered work" means either the unmodified Program or a work based
+on the Program.
+
+  To "propagate" a work means to do anything with it that, without
+permission, would make you directly or secondarily liable for
+infringement under applicable copyright law, except executing it on a
+computer or modifying a private copy.  Propagation includes copying,
+distribution (with or without modification), making available to the
+public, and in some countries other activities as well.
+
+  To "convey" a work means any kind of propagation that enables other
+parties to make or receive copies.  Mere interaction with a user through
+a computer network, with no transfer of a copy, is not conveying.
+
+  An interactive user interface displays "Appropriate Legal Notices"
+to the extent that it includes a convenient and prominently visible
+feature that (1) displays an appropriate copyright notice, and (2)
+tells the user that there is no warranty for the work (except to the
+extent that warranties are provided), that licensees may convey the
+work under this License, and how to view a copy of this License.  If
+the interface presents a list of user commands or options, such as a
+menu, a prominent item in the list meets this criterion.
+
+  1. Source Code.
+
+  The "source code" for a work means the preferred form of the work
+for making modifications to it.  "Object code" means any non-source
+form of a work.
+
+  A "Standard Interface" means an interface that either is an official
+standard defined by a recognized standards body, or, in the case of
+interfaces specified for a particular programming language, one that
+is widely used among developers working in that language.
+
+  The "System Libraries" of an executable work include anything, other
+than the work as a whole, that (a) is included in the normal form of
+packaging a Major Component, but which is not part of that Major
+Component, and (b) serves only to enable use of the work with that
+Major Component, or to implement a Standard Interface for which an
+implementation is available to the public in source code form.  A
+"Major Component", in this context, means a major essential component
+(kernel, window system, and so on) of the specific operating system
+(if any) on which the executable work runs, or a compiler used to
+produce the work, or an object code interpreter used to run it.
+
+  The "Corresponding Source" for a work in object code form means all
+the source code needed to generate, install, and (for an executable
+work) run the object code and to modify the work, including scripts to
+control those activities.  However, it does not include the work's
+System Libraries, or general-purpose tools or generally available free
+programs which are used unmodified in performing those activities but
+which are not part of the work.  For example, Corresponding Source
+includes interface definition files associated with source files for
+the work, and the source code for shared libraries and dynamically
+linked subprograms that the work is specifically designed to require,
+such as by intimate data communication or control flow between those
+subprograms and other parts of the work.
+
+  The Corresponding Source need not include anything that users
+can regenerate automatically from other parts of the Corresponding
+Source.
+
+  The Corresponding Source for a work in source code form is that
+same work.
+
+  2. Basic Permissions.
+
+  All rights granted under this License are granted for the term of
+copyright on the Program, and are irrevocable provided the stated
+conditions are met.  This License explicitly affirms your unlimited
+permission to run the unmodified Program.  The output from running a
+covered work is covered by this License only if the output, given its
+content, constitutes a covered work.  This License acknowledges your
+rights of fair use or other equivalent, as provided by copyright law.
+
+  You may make, run and propagate covered works that you do not
+convey, without conditions so long as your license otherwise remains
+in force.  You may convey covered works to others for the sole purpose
+of having them make modifications exclusively for you, or provide you
+with facilities for running those works, provided that you comply with
+the terms of this License in conveying all material for which you do
+not control copyright.  Those thus making or running the covered works
+for you must do so exclusively on your behalf, under your direction
+and control, on terms that prohibit them from making any copies of
+your copyrighted material outside their relationship with you.
+
+  Conveying under any other circumstances is permitted solely under
+the conditions stated below.  Sublicensing is not allowed; section 10
+makes it unnecessary.
+
+  3. Protecting Users' Legal Rights From Anti-Circumvention Law.
+
+  No covered work shall be deemed part of an effective technological
+measure under any applicable law fulfilling obligations under article
+11 of the WIPO copyright treaty adopted on 20 December 1996, or
+similar laws prohibiting or restricting circumvention of such
+measures.
+
+  When you convey a covered work, you waive any legal power to forbid
+circumvention of technological measures to the extent such circumvention
+is effected by exercising rights under this License with respect to
+the covered work, and you disclaim any intention to limit operation or
+modification of the work as a means of enforcing, against the work's
+users, your or third parties' legal rights to forbid circumvention of
+technological measures.
+
+  4. Conveying Verbatim Copies.
+
+  You may convey verbatim copies of the Program's source code as you
+receive it, in any medium, provided that you conspicuously and
+appropriately publish on each copy an appropriate copyright notice;
+keep intact all notices stating that this License and any
+non-permissive terms added in accord with section 7 apply to the code;
+keep intact all notices of the absence of any warranty; and give all
+recipients a copy of this License along with the Program.
+
+  You may charge any price or no price for each copy that you convey,
+and you may offer support or warranty protection for a fee.
+
+  5. Conveying Modified Source Versions.
+
+  You may convey a work based on the Program, or the modifications to
+produce it from the Program, in the form of source code under the
+terms of section 4, provided that you also meet all of these conditions:
+
+    a) The work must carry prominent notices stating that you modified
+    it, and giving a relevant date.
+
+    b) The work must carry prominent notices stating that it is
+    released under this License and any conditions added under section
+    7.  This requirement modifies the requirement in section 4 to
+    "keep intact all notices".
+
+    c) You must license the entire work, as a whole, under this
+    License to anyone who comes into possession of a copy.  This
+    License will therefore apply, along with any applicable section 7
+    additional terms, to the whole of the work, and all its parts,
+    regardless of how they are packaged.  This License gives no
+    permission to license the work in any other way, but it does not
+    invalidate such permission if you have separately received it.
+
+    d) If the work has interactive user interfaces, each must display
+    Appropriate Legal Notices; however, if the Program has interactive
+    interfaces that do not display Appropriate Legal Notices, your
+    work need not make them do so.
+
+  A compilation of a covered work with other separate and independent
+works, which are not by their nature extensions of the covered work,
+and which are not combined with it such as to form a larger program,
+in or on a volume of a storage or distribution medium, is called an
+"aggregate" if the compilation and its resulting copyright are not
+used to limit the access or legal rights of the compilation's users
+beyond what the individual works permit.  Inclusion of a covered work
+in an aggregate does not cause this License to apply to the other
+parts of the aggregate.
+
+  6. Conveying Non-Source Forms.
+
+  You may convey a covered work in object code form under the terms
+of sections 4 and 5, provided that you also convey the
+machine-readable Corresponding Source under the terms of this License,
+in one of these ways:
+
+    a) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by the
+    Corresponding Source fixed on a durable physical medium
+    customarily used for software interchange.
+
+    b) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by a
+    written offer, valid for at least three years and valid for as
+    long as you offer spare parts or customer support for that product
+    model, to give anyone who possesses the object code either (1) a
+    copy of the Corresponding Source for all the software in the
+    product that is covered by this License, on a durable physical
+    medium customarily used for software interchange, for a price no
+    more than your reasonable cost of physically performing this
+    conveying of source, or (2) access to copy the
+    Corresponding Source from a network server at no charge.
+
+    c) Convey individual copies of the object code with a copy of the
+    written offer to provide the Corresponding Source.  This
+    alternative is allowed only occasionally and noncommercially, and
+    only if you received the object code with such an offer, in accord
+    with subsection 6b.
+
+    d) Convey the object code by offering access from a designated
+    place (gratis or for a charge), and offer equivalent access to the
+    Corresponding Source in the same way through the same place at no
+    further charge.  You need not require recipients to copy the
+    Corresponding Source along with the object code.  If the place to
+    copy the object code is a network server, the Corresponding Source
+    may be on a different server (operated by you or a third party)
+    that supports equivalent copying facilities, provided you maintain
+    clear directions next to the object code saying where to find the
+    Corresponding Source.  Regardless of what server hosts the
+    Corresponding Source, you remain obligated to ensure that it is
+    available for as long as needed to satisfy these requirements.
+
+    e) Convey the object code using peer-to-peer transmission, provided
+    you inform other peers where the object code and Corresponding
+    Source of the work are being offered to the general public at no
+    charge under subsection 6d.
+
+  A separable portion of the object code, whose source code is excluded
+from the Corresponding Source as a System Library, need not be
+included in conveying the object code work.
+
+  A "User Product" is either (1) a "consumer product", which means any
+tangible personal property which is normally used for personal, family,
+or household purposes, or (2) anything designed or sold for incorporation
+into a dwelling.  In determining whether a product is a consumer product,
+doubtful cases shall be resolved in favor of coverage.  For a particular
+product received by a particular user, "normally used" refers to a
+typical or common use of that class of product, regardless of the status
+of the particular user or of the way in which the particular user
+actually uses, or expects or is expected to use, the product.  A product
+is a consumer product regardless of whether the product has substantial
+commercial, industrial or non-consumer uses, unless such uses represent
+the only significant mode of use of the product.
+
+  "Installation Information" for a User Product means any methods,
+procedures, authorization keys, or other information required to install
+and execute modified versions of a covered work in that User Product from
+a modified version of its Corresponding Source.  The information must
+suffice to ensure that the continued functioning of the modified object
+code is in no case prevented or interfered with solely because
+modification has been made.
+
+  If you convey an object code work under this section in, or with, or
+specifically for use in, a User Product, and the conveying occurs as
+part of a transaction in which the right of possession and use of the
+User Product is transferred to the recipient in perpetuity or for a
+fixed term (regardless of how the transaction is characterized), the
+Corresponding Source conveyed under this section must be accompanied
+by the Installation Information.  But this requirement does not apply
+if neither you nor any third party retains the ability to install
+modified object code on the User Product (for example, the work has
+been installed in ROM).
+
+  The requirement to provide Installation Information does not include a
+requirement to continue to provide support service, warranty, or updates
+for a work that has been modified or installed by the recipient, or for
+the User Product in which it has been modified or installed.  Access to a
+network may be denied when the modification itself materially and
+adversely affects the operation of the network or violates the rules and
+protocols for communication across the network.
+
+  Corresponding Source conveyed, and Installation Information provided,
+in accord with this section must be in a format that is publicly
+documented (and with an implementation available to the public in
+source code form), and must require no special password or key for
+unpacking, reading or copying.
+
+  7. Additional Terms.
+
+  "Additional permissions" are terms that supplement the terms of this
+License by making exceptions from one or more of its conditions.
+Additional permissions that are applicable to the entire Program shall
+be treated as though they were included in this License, to the extent
+that they are valid under applicable law.  If additional permissions
+apply only to part of the Program, that part may be used separately
+under those permissions, but the entire Program remains governed by
+this License without regard to the additional permissions.
+
+  When you convey a copy of a covered work, you may at your option
+remove any additional permissions from that copy, or from any part of
+it.  (Additional permissions may be written to require their own
+removal in certain cases when you modify the work.)  You may place
+additional permissions on material, added by you to a covered work,
+for which you have or can give appropriate copyright permission.
+
+  Notwithstanding any other provision of this License, for material you
+add to a covered work, you may (if authorized by the copyright holders of
+that material) supplement the terms of this License with terms:
+
+    a) Disclaiming warranty or limiting liability differently from the
+    terms of sections 15 and 16 of this License; or
+
+    b) Requiring preservation of specified reasonable legal notices or
+    author attributions in that material or in the Appropriate Legal
+    Notices displayed by works containing it; or
+
+    c) Prohibiting misrepresentation of the origin of that material, or
+    requiring that modified versions of such material be marked in
+    reasonable ways as different from the original version; or
+
+    d) Limiting the use for publicity purposes of names of licensors or
+    authors of the material; or
+
+    e) Declining to grant rights under trademark law for use of some
+    trade names, trademarks, or service marks; or
+
+    f) Requiring indemnification of licensors and authors of that
+    material by anyone who conveys the material (or modified versions of
+    it) with contractual assumptions of liability to the recipient, for
+    any liability that these contractual assumptions directly impose on
+    those licensors and authors.
+
+  All other non-permissive additional terms are considered "further
+restrictions" within the meaning of section 10.  If the Program as you
+received it, or any part of it, contains a notice stating that it is
+governed by this License along with a term that is a further
+restriction, you may remove that term.  If a license document contains
+a further restriction but permits relicensing or conveying under this
+License, you may add to a covered work material governed by the terms
+of that license document, provided that the further restriction does
+not survive such relicensing or conveying.
+
+  If you add terms to a covered work in accord with this section, you
+must place, in the relevant source files, a statement of the
+additional terms that apply to those files, or a notice indicating
+where to find the applicable terms.
+
+  Additional terms, permissive or non-permissive, may be stated in the
+form of a separately written license, or stated as exceptions;
+the above requirements apply either way.
+
+  8. Termination.
+
+  You may not propagate or modify a covered work except as expressly
+provided under this License.  Any attempt otherwise to propagate or
+modify it is void, and will automatically terminate your rights under
+this License (including any patent licenses granted under the third
+paragraph of section 11).
+
+  However, if you cease all violation of this License, then your
+license from a particular copyright holder is reinstated (a)
+provisionally, unless and until the copyright holder explicitly and
+finally terminates your license, and (b) permanently, if the copyright
+holder fails to notify you of the violation by some reasonable means
+prior to 60 days after the cessation.
+
+  Moreover, your license from a particular copyright holder is
+reinstated permanently if the copyright holder notifies you of the
+violation by some reasonable means, this is the first time you have
+received notice of violation of this License (for any work) from that
+copyright holder, and you cure the violation prior to 30 days after
+your receipt of the notice.
+
+  Termination of your rights under this section does not terminate the
+licenses of parties who have received copies or rights from you under
+this License.  If your rights have been terminated and not permanently
+reinstated, you do not qualify to receive new licenses for the same
+material under section 10.
+
+  9. Acceptance Not Required for Having Copies.
+
+  You are not required to accept this License in order to receive or
+run a copy of the Program.  Ancillary propagation of a covered work
+occurring solely as a consequence of using peer-to-peer transmission
+to receive a copy likewise does not require acceptance.  However,
+nothing other than this License grants you permission to propagate or
+modify any covered work.  These actions infringe copyright if you do
+not accept this License.  Therefore, by modifying or propagating a
+covered work, you indicate your acceptance of this License to do so.
+
+  10. Automatic Licensing of Downstream Recipients.
+
+  Each time you convey a covered work, the recipient automatically
+receives a license from the original licensors, to run, modify and
+propagate that work, subject to this License.  You are not responsible
+for enforcing compliance by third parties with this License.
+
+  An "entity transaction" is a transaction transferring control of an
+organization, or substantially all assets of one, or subdividing an
+organization, or merging organizations.  If propagation of a covered
+work results from an entity transaction, each party to that
+transaction who receives a copy of the work also receives whatever
+licenses to the work the party's predecessor in interest had or could
+give under the previous paragraph, plus a right to possession of the
+Corresponding Source of the work from the predecessor in interest, if
+the predecessor has it or can get it with reasonable efforts.
+
+  You may not impose any further restrictions on the exercise of the
+rights granted or affirmed under this License.  For example, you may
+not impose a license fee, royalty, or other charge for exercise of
+rights granted under this License, and you may not initiate litigation
+(including a cross-claim or counterclaim in a lawsuit) alleging that
+any patent claim is infringed by making, using, selling, offering for
+sale, or importing the Program or any portion of it.
+
+  11. Patents.
+
+  A "contributor" is a copyright holder who authorizes use under this
+License of the Program or a work on which the Program is based.  The
+work thus licensed is called the contributor's "contributor version".
+
+  A contributor's "essential patent claims" are all patent claims
+owned or controlled by the contributor, whether already acquired or
+hereafter acquired, that would be infringed by some manner, permitted
+by this License, of making, using, or selling its contributor version,
+but do not include claims that would be infringed only as a
+consequence of further modification of the contributor version.  For
+purposes of this definition, "control" includes the right to grant
+patent sublicenses in a manner consistent with the requirements of
+this License.
+
+  Each contributor grants you a non-exclusive, worldwide, royalty-free
+patent license under the contributor's essential patent claims, to
+make, use, sell, offer for sale, import and otherwise run, modify and
+propagate the contents of its contributor version.
+
+  In the following three paragraphs, a "patent license" is any express
+agreement or commitment, however denominated, not to enforce a patent
+(such as an express permission to practice a patent or covenant not to
+sue for patent infringement).  To "grant" such a patent license to a
+party means to make such an agreement or commitment not to enforce a
+patent against the party.
+
+  If you convey a covered work, knowingly relying on a patent license,
+and the Corresponding Source of the work is not available for anyone
+to copy, free of charge and under the terms of this License, through a
+publicly available network server or other readily accessible means,
+then you must either (1) cause the Corresponding Source to be so
+available, or (2) arrange to deprive yourself of the benefit of the
+patent license for this particular work, or (3) arrange, in a manner
+consistent with the requirements of this License, to extend the patent
+license to downstream recipients.  "Knowingly relying" means you have
+actual knowledge that, but for the patent license, your conveying the
+covered work in a country, or your recipient's use of the covered work
+in a country, would infringe one or more identifiable patents in that
+country that you have reason to believe are valid.
+
+  If, pursuant to or in connection with a single transaction or
+arrangement, you convey, or propagate by procuring conveyance of, a
+covered work, and grant a patent license to some of the parties
+receiving the covered work authorizing them to use, propagate, modify
+or convey a specific copy of the covered work, then the patent license
+you grant is automatically extended to all recipients of the covered
+work and works based on it.
+
+  A patent license is "discriminatory" if it does not include within
+the scope of its coverage, prohibits the exercise of, or is
+conditioned on the non-exercise of one or more of the rights that are
+specifically granted under this License.  You may not convey a covered
+work if you are a party to an arrangement with a third party that is
+in the business of distributing software, under which you make payment
+to the third party based on the extent of your activity of conveying
+the work, and under which the third party grants, to any of the
+parties who would receive the covered work from you, a discriminatory
+patent license (a) in connection with copies of the covered work
+conveyed by you (or copies made from those copies), or (b) primarily
+for and in connection with specific products or compilations that
+contain the covered work, unless you entered into that arrangement,
+or that patent license was granted, prior to 28 March 2007.
+
+  Nothing in this License shall be construed as excluding or limiting
+any implied license or other defenses to infringement that may
+otherwise be available to you under applicable patent law.
+
+  12. No Surrender of Others' Freedom.
+
+  If conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License.  If you cannot convey a
+covered work so as to satisfy simultaneously your obligations under this
+License and any other pertinent obligations, then as a consequence you may
+not convey it at all.  For example, if you agree to terms that obligate you
+to collect a royalty for further conveying from those to whom you convey
+the Program, the only way you could satisfy both those terms and this
+License would be to refrain entirely from conveying the Program.
+
+  13. Use with the GNU Affero General Public License.
+
+  Notwithstanding any other provision of this License, you have
+permission to link or combine any covered work with a work licensed
+under version 3 of the GNU Affero General Public License into a single
+combined work, and to convey the resulting work.  The terms of this
+License will continue to apply to the part which is the covered work,
+but the special requirements of the GNU Affero General Public License,
+section 13, concerning interaction through a network will apply to the
+combination as such.
+
+  14. Revised Versions of this License.
+
+  The Free Software Foundation may publish revised and/or new versions of
+the GNU General Public License from time to time.  Such new versions will
+be similar in spirit to the present version, but may differ in detail to
+address new problems or concerns.
+
+  Each version is given a distinguishing version number.  If the
+Program specifies that a certain numbered version of the GNU General
+Public License "or any later version" applies to it, you have the
+option of following the terms and conditions either of that numbered
+version or of any later version published by the Free Software
+Foundation.  If the Program does not specify a version number of the
+GNU General Public License, you may choose any version ever published
+by the Free Software Foundation.
+
+  If the Program specifies that a proxy can decide which future
+versions of the GNU General Public License can be used, that proxy's
+public statement of acceptance of a version permanently authorizes you
+to choose that version for the Program.
+
+  Later license versions may give you additional or different
+permissions.  However, no additional obligations are imposed on any
+author or copyright holder as a result of your choosing to follow a
+later version.
+
+  15. Disclaimer of Warranty.
+
+  THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
+APPLICABLE LAW.  EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
+HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
+OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
+THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+PURPOSE.  THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
+IS WITH YOU.  SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
+ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
+
+  16. Limitation of Liability.
+
+  IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
+THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
+GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
+USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
+DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
+PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
+EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
+SUCH DAMAGES.
+
+  17. Interpretation of Sections 15 and 16.
+
+  If the disclaimer of warranty and limitation of liability provided
+above cannot be given local legal effect according to their terms,
+reviewing courts shall apply local law that most closely approximates
+an absolute waiver of all civil liability in connection with the
+Program, unless a warranty or assumption of liability accompanies a
+copy of the Program in return for a fee.
+
+                     END OF TERMS AND CONDITIONS
+
+            How to Apply These Terms to Your New Programs
+
+  If you develop a new program, and you want it to be of the greatest
+possible use to the public, the best way to achieve this is to make it
+free software which everyone can redistribute and change under these terms.
+
+  To do so, attach the following notices to the program.  It is safest
+to attach them to the start of each source file to most effectively
+state the exclusion of warranty; and each file should have at least
+the "copyright" line and a pointer to where the full notice is found.
+
+    <one line to give the program's name and a brief idea of what it does.>
+    Copyright (C) <year>  <name of author>
+
+    This program is free software: you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+Also add information on how to contact you by electronic and paper mail.
+
+  If the program does terminal interaction, make it output a short
+notice like this when it starts in an interactive mode:
+
+    <program>  Copyright (C) <year>  <name of author>
+    This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
+    This is free software, and you are welcome to redistribute it
+    under certain conditions; type `show c' for details.
+
+The hypothetical commands `show w' and `show c' should show the appropriate
+parts of the General Public License.  Of course, your program's commands
+might be different; for a GUI interface, you would use an "about box".
+
+  You should also get your employer (if you work as a programmer) or school,
+if any, to sign a "copyright disclaimer" for the program, if necessary.
+For more information on this, and how to apply and follow the GNU GPL, see
+<https://www.gnu.org/licenses/>.
+
+  The GNU General Public License does not permit incorporating your program
+into proprietary programs.  If your program is a subroutine library, you
+may consider it more useful to permit linking proprietary applications with
+the library.  If this is what you want to do, use the GNU Lesser General
+Public License instead of this License.  But first, please read
+<https://www.gnu.org/licenses/why-not-lgpl.html>.

backend/headless/fcbh/checkpoint_pickle.py

@@ -0,0 +1,13 @@
+import pickle
+
+load = pickle.load
+
+class Empty:
+    pass
+
+class Unpickler(pickle.Unpickler):
+    def find_class(self, module, name):
+        #TODO: safe unpickle
+        if module.startswith("pytorch_lightning"):
+            return Empty
+        return super().find_class(module, name)

backend/headless/fcbh/cldm/cldm.py

@@ -0,0 +1,311 @@
+#taken from: https://github.com/lllyasviel/ControlNet
+#and modified
+
+import torch
+import torch as th
+import torch.nn as nn
+
+from ..ldm.modules.diffusionmodules.util import (
+    zero_module,
+    timestep_embedding,
+)
+
+from ..ldm.modules.attention import SpatialTransformer
+from ..ldm.modules.diffusionmodules.openaimodel import UNetModel, TimestepEmbedSequential, ResBlock, Downsample
+from ..ldm.util import exists
+import fcbh.ops
+
+class ControlledUnetModel(UNetModel):
+    #implemented in the ldm unet
+    pass
+
+class ControlNet(nn.Module):
+    def __init__(
+        self,
+        image_size,
+        in_channels,
+        model_channels,
+        hint_channels,
+        num_res_blocks,
+        dropout=0,
+        channel_mult=(1, 2, 4, 8),
+        conv_resample=True,
+        dims=2,
+        num_classes=None,
+        use_checkpoint=False,
+        dtype=torch.float32,
+        num_heads=-1,
+        num_head_channels=-1,
+        num_heads_upsample=-1,
+        use_scale_shift_norm=False,
+        resblock_updown=False,
+        use_new_attention_order=False,
+        use_spatial_transformer=False,    # custom transformer support
+        transformer_depth=1,              # custom transformer support
+        context_dim=None,                 # custom transformer support
+        n_embed=None,                     # custom support for prediction of discrete ids into codebook of first stage vq model
+        legacy=True,
+        disable_self_attentions=None,
+        num_attention_blocks=None,
+        disable_middle_self_attn=False,
+        use_linear_in_transformer=False,
+        adm_in_channels=None,
+        transformer_depth_middle=None,
+        transformer_depth_output=None,
+        device=None,
+        operations=fcbh.ops,
+    ):
+        super().__init__()
+        assert use_spatial_transformer == True, "use_spatial_transformer has to be true"
+        if use_spatial_transformer:
+            assert context_dim is not None, 'Fool!! You forgot to include the dimension of your cross-attention conditioning...'
+
+        if context_dim is not None:
+            assert use_spatial_transformer, 'Fool!! You forgot to use the spatial transformer for your cross-attention conditioning...'
+            # from omegaconf.listconfig import ListConfig
+            # if type(context_dim) == ListConfig:
+            #     context_dim = list(context_dim)
+
+        if num_heads_upsample == -1:
+            num_heads_upsample = num_heads
+
+        if num_heads == -1:
+            assert num_head_channels != -1, 'Either num_heads or num_head_channels has to be set'
+
+        if num_head_channels == -1:
+            assert num_heads != -1, 'Either num_heads or num_head_channels has to be set'
+
+        self.dims = dims
+        self.image_size = image_size
+        self.in_channels = in_channels
+        self.model_channels = model_channels
+
+        if isinstance(num_res_blocks, int):
+            self.num_res_blocks = len(channel_mult) * [num_res_blocks]
+        else:
+            if len(num_res_blocks) != len(channel_mult):
+                raise ValueError("provide num_res_blocks either as an int (globally constant) or "
+                                 "as a list/tuple (per-level) with the same length as channel_mult")
+            self.num_res_blocks = num_res_blocks
+
+        if disable_self_attentions is not None:
+            # should be a list of booleans, indicating whether to disable self-attention in TransformerBlocks or not
+            assert len(disable_self_attentions) == len(channel_mult)
+        if num_attention_blocks is not None:
+            assert len(num_attention_blocks) == len(self.num_res_blocks)
+            assert all(map(lambda i: self.num_res_blocks[i] >= num_attention_blocks[i], range(len(num_attention_blocks))))
+
+        transformer_depth = transformer_depth[:]
+
+        self.dropout = dropout
+        self.channel_mult = channel_mult
+        self.conv_resample = conv_resample
+        self.num_classes = num_classes
+        self.use_checkpoint = use_checkpoint
+        self.dtype = dtype
+        self.num_heads = num_heads
+        self.num_head_channels = num_head_channels
+        self.num_heads_upsample = num_heads_upsample
+        self.predict_codebook_ids = n_embed is not None
+
+        time_embed_dim = model_channels * 4
+        self.time_embed = nn.Sequential(
+            operations.Linear(model_channels, time_embed_dim, dtype=self.dtype, device=device),
+            nn.SiLU(),
+            operations.Linear(time_embed_dim, time_embed_dim, dtype=self.dtype, device=device),
+        )
+
+        if self.num_classes is not None:
+            if isinstance(self.num_classes, int):
+                self.label_emb = nn.Embedding(num_classes, time_embed_dim)
+            elif self.num_classes == "continuous":
+                print("setting up linear c_adm embedding layer")
+                self.label_emb = nn.Linear(1, time_embed_dim)
+            elif self.num_classes == "sequential":
+                assert adm_in_channels is not None
+                self.label_emb = nn.Sequential(
+                    nn.Sequential(
+                        operations.Linear(adm_in_channels, time_embed_dim, dtype=self.dtype, device=device),
+                        nn.SiLU(),
+                        operations.Linear(time_embed_dim, time_embed_dim, dtype=self.dtype, device=device),
+                    )
+                )
+            else:
+                raise ValueError()
+
+        self.input_blocks = nn.ModuleList(
+            [
+                TimestepEmbedSequential(
+                    operations.conv_nd(dims, in_channels, model_channels, 3, padding=1, dtype=self.dtype, device=device)
+                )
+            ]
+        )
+        self.zero_convs = nn.ModuleList([self.make_zero_conv(model_channels, operations=operations)])
+
+        self.input_hint_block = TimestepEmbedSequential(
+                    operations.conv_nd(dims, hint_channels, 16, 3, padding=1),
+                    nn.SiLU(),
+                    operations.conv_nd(dims, 16, 16, 3, padding=1),
+                    nn.SiLU(),
+                    operations.conv_nd(dims, 16, 32, 3, padding=1, stride=2),
+                    nn.SiLU(),
+                    operations.conv_nd(dims, 32, 32, 3, padding=1),
+                    nn.SiLU(),
+                    operations.conv_nd(dims, 32, 96, 3, padding=1, stride=2),
+                    nn.SiLU(),
+                    operations.conv_nd(dims, 96, 96, 3, padding=1),
+                    nn.SiLU(),
+                    operations.conv_nd(dims, 96, 256, 3, padding=1, stride=2),
+                    nn.SiLU(),
+                    zero_module(operations.conv_nd(dims, 256, model_channels, 3, padding=1))
+        )
+
+        self._feature_size = model_channels
+        input_block_chans = [model_channels]
+        ch = model_channels
+        ds = 1
+        for level, mult in enumerate(channel_mult):
+            for nr in range(self.num_res_blocks[level]):
+                layers = [
+                    ResBlock(
+                        ch,
+                        time_embed_dim,
+                        dropout,
+                        out_channels=mult * model_channels,
+                        dims=dims,
+                        use_checkpoint=use_checkpoint,
+                        use_scale_shift_norm=use_scale_shift_norm,
+                        dtype=self.dtype,
+                        device=device,
+                        operations=operations,
+                    )
+                ]
+                ch = mult * model_channels
+                num_transformers = transformer_depth.pop(0)
+                if num_transformers > 0:
+                    if num_head_channels == -1:
+                        dim_head = ch // num_heads
+                    else:
+                        num_heads = ch // num_head_channels
+                        dim_head = num_head_channels
+                    if legacy:
+                        #num_heads = 1
+                        dim_head = ch // num_heads if use_spatial_transformer else num_head_channels
+                    if exists(disable_self_attentions):
+                        disabled_sa = disable_self_attentions[level]
+                    else:
+                        disabled_sa = False
+
+                    if not exists(num_attention_blocks) or nr < num_attention_blocks[level]:
+                        layers.append(
+                            SpatialTransformer(
+                                ch, num_heads, dim_head, depth=num_transformers, context_dim=context_dim,
+                                disable_self_attn=disabled_sa, use_linear=use_linear_in_transformer,
+                                use_checkpoint=use_checkpoint, dtype=self.dtype, device=device, operations=operations
+                            )
+                        )
+                self.input_blocks.append(TimestepEmbedSequential(*layers))
+                self.zero_convs.append(self.make_zero_conv(ch, operations=operations))
+                self._feature_size += ch
+                input_block_chans.append(ch)
+            if level != len(channel_mult) - 1:
+                out_ch = ch
+                self.input_blocks.append(
+                    TimestepEmbedSequential(
+                        ResBlock(
+                            ch,
+                            time_embed_dim,
+                            dropout,
+                            out_channels=out_ch,
+                            dims=dims,
+                            use_checkpoint=use_checkpoint,
+                            use_scale_shift_norm=use_scale_shift_norm,
+                            down=True,
+                            dtype=self.dtype,
+                            device=device,
+                            operations=operations
+                        )
+                        if resblock_updown
+                        else Downsample(
+                            ch, conv_resample, dims=dims, out_channels=out_ch, dtype=self.dtype, device=device, operations=operations
+                        )
+                    )
+                )
+                ch = out_ch
+                input_block_chans.append(ch)
+                self.zero_convs.append(self.make_zero_conv(ch, operations=operations))
+                ds *= 2
+                self._feature_size += ch
+
+        if num_head_channels == -1:
+            dim_head = ch // num_heads
+        else:
+            num_heads = ch // num_head_channels
+            dim_head = num_head_channels
+        if legacy:
+            #num_heads = 1
+            dim_head = ch // num_heads if use_spatial_transformer else num_head_channels
+        mid_block = [
+            ResBlock(
+                ch,
+                time_embed_dim,
+                dropout,
+                dims=dims,
+                use_checkpoint=use_checkpoint,
+                use_scale_shift_norm=use_scale_shift_norm,
+                dtype=self.dtype,
+                device=device,
+                operations=operations
+            )]
+        if transformer_depth_middle >= 0:
+            mid_block += [SpatialTransformer(  # always uses a self-attn
+                            ch, num_heads, dim_head, depth=transformer_depth_middle, context_dim=context_dim,
+                            disable_self_attn=disable_middle_self_attn, use_linear=use_linear_in_transformer,
+                            use_checkpoint=use_checkpoint, dtype=self.dtype, device=device, operations=operations
+                        ),
+            ResBlock(
+                ch,
+                time_embed_dim,
+                dropout,
+                dims=dims,
+                use_checkpoint=use_checkpoint,
+                use_scale_shift_norm=use_scale_shift_norm,
+                dtype=self.dtype,
+                device=device,
+                operations=operations
+            )]
+        self.middle_block = TimestepEmbedSequential(*mid_block)
+        self.middle_block_out = self.make_zero_conv(ch, operations=operations)
+        self._feature_size += ch
+
+    def make_zero_conv(self, channels, operations=None):
+        return TimestepEmbedSequential(zero_module(operations.conv_nd(self.dims, channels, channels, 1, padding=0)))
+
+    def forward(self, x, hint, timesteps, context, y=None, **kwargs):
+        t_emb = timestep_embedding(timesteps, self.model_channels, repeat_only=False).to(self.dtype)
+        emb = self.time_embed(t_emb)
+
+        guided_hint = self.input_hint_block(hint, emb, context)
+
+        outs = []
+
+        hs = []
+        if self.num_classes is not None:
+            assert y.shape[0] == x.shape[0]
+            emb = emb + self.label_emb(y)
+
+        h = x.type(self.dtype)
+        for module, zero_conv in zip(self.input_blocks, self.zero_convs):
+            if guided_hint is not None:
+                h = module(h, emb, context)
+                h += guided_hint
+                guided_hint = None
+            else:
+                h = module(h, emb, context)
+            outs.append(zero_conv(h, emb, context))
+
+        h = self.middle_block(h, emb, context)
+        outs.append(self.middle_block_out(h, emb, context))
+
+        return outs
+

backend/headless/fcbh/cli_args.py

@@ -0,0 +1,111 @@
+import argparse
+import enum
+import fcbh.options
+
+class EnumAction(argparse.Action):
+    """
+    Argparse action for handling Enums
+    """
+    def __init__(self, **kwargs):
+        # Pop off the type value
+        enum_type = kwargs.pop("type", None)
+
+        # Ensure an Enum subclass is provided
+        if enum_type is None:
+            raise ValueError("type must be assigned an Enum when using EnumAction")
+        if not issubclass(enum_type, enum.Enum):
+            raise TypeError("type must be an Enum when using EnumAction")
+
+        # Generate choices from the Enum
+        choices = tuple(e.value for e in enum_type)
+        kwargs.setdefault("choices", choices)
+        kwargs.setdefault("metavar", f"[{','.join(list(choices))}]")
+
+        super(EnumAction, self).__init__(**kwargs)
+
+        self._enum = enum_type
+
+    def __call__(self, parser, namespace, values, option_string=None):
+        # Convert value back into an Enum
+        value = self._enum(values)
+        setattr(namespace, self.dest, value)
+
+
+parser = argparse.ArgumentParser()
+
+parser.add_argument("--listen", type=str, default="127.0.0.1", metavar="IP", nargs="?", const="0.0.0.0", help="Specify the IP address to listen on (default: 127.0.0.1). If --listen is provided without an argument, it defaults to 0.0.0.0. (listens on all)")
+parser.add_argument("--port", type=int, default=8188, help="Set the listen port.")
+parser.add_argument("--enable-cors-header", type=str, default=None, metavar="ORIGIN", nargs="?", const="*", help="Enable CORS (Cross-Origin Resource Sharing) with optional origin or allow all with default '*'.")
+parser.add_argument("--max-upload-size", type=float, default=100, help="Set the maximum upload size in MB.")
+
+parser.add_argument("--extra-model-paths-config", type=str, default=None, metavar="PATH", nargs='+', action='append', help="Load one or more extra_model_paths.yaml files.")
+parser.add_argument("--output-directory", type=str, default=None, help="Set the fcbh_backend output directory.")
+parser.add_argument("--temp-directory", type=str, default=None, help="Set the fcbh_backend temp directory (default is in the fcbh_backend directory).")
+parser.add_argument("--input-directory", type=str, default=None, help="Set the fcbh_backend input directory.")
+parser.add_argument("--auto-launch", action="store_true", help="Automatically launch fcbh_backend in the default browser.")
+parser.add_argument("--disable-auto-launch", action="store_true", help="Disable auto launching the browser.")
+parser.add_argument("--cuda-device", type=int, default=None, metavar="DEVICE_ID", help="Set the id of the cuda device this instance will use.")
+cm_group = parser.add_mutually_exclusive_group()
+cm_group.add_argument("--cuda-malloc", action="store_true", help="Enable cudaMallocAsync (enabled by default for torch 2.0 and up).")
+cm_group.add_argument("--disable-cuda-malloc", action="store_true", help="Disable cudaMallocAsync.")
+
+parser.add_argument("--dont-upcast-attention", action="store_true", help="Disable upcasting of attention. Can boost speed but increase the chances of black images.")
+
+fp_group = parser.add_mutually_exclusive_group()
+fp_group.add_argument("--force-fp32", action="store_true", help="Force fp32 (If this makes your GPU work better please report it).")
+fp_group.add_argument("--force-fp16", action="store_true", help="Force fp16.")
+
+parser.add_argument("--bf16-unet", action="store_true", help="Run the UNET in bf16. This should only be used for testing stuff.")
+
+fpvae_group = parser.add_mutually_exclusive_group()
+fpvae_group.add_argument("--fp16-vae", action="store_true", help="Run the VAE in fp16, might cause black images.")
+fpvae_group.add_argument("--fp32-vae", action="store_true", help="Run the VAE in full precision fp32.")
+fpvae_group.add_argument("--bf16-vae", action="store_true", help="Run the VAE in bf16.")
+
+parser.add_argument("--directml", type=int, nargs="?", metavar="DIRECTML_DEVICE", const=-1, help="Use torch-directml.")
+
+parser.add_argument("--disable-ipex-optimize", action="store_true", help="Disables ipex.optimize when loading models with Intel GPUs.")
+
+class LatentPreviewMethod(enum.Enum):
+    NoPreviews = "none"
+    Auto = "auto"
+    Latent2RGB = "latent2rgb"
+    TAESD = "taesd"
+
+parser.add_argument("--preview-method", type=LatentPreviewMethod, default=LatentPreviewMethod.NoPreviews, help="Default preview method for sampler nodes.", action=EnumAction)
+
+attn_group = parser.add_mutually_exclusive_group()
+attn_group.add_argument("--use-split-cross-attention", action="store_true", help="Use the split cross attention optimization. Ignored when xformers is used.")
+attn_group.add_argument("--use-quad-cross-attention", action="store_true", help="Use the sub-quadratic cross attention optimization . Ignored when xformers is used.")
+attn_group.add_argument("--use-pytorch-cross-attention", action="store_true", help="Use the new pytorch 2.0 cross attention function.")
+
+parser.add_argument("--disable-xformers", action="store_true", help="Disable xformers.")
+
+vram_group = parser.add_mutually_exclusive_group()
+vram_group.add_argument("--gpu-only", action="store_true", help="Store and run everything (text encoders/CLIP models, etc... on the GPU).")
+vram_group.add_argument("--highvram", action="store_true", help="By default models will be unloaded to CPU memory after being used. This option keeps them in GPU memory.")
+vram_group.add_argument("--normalvram", action="store_true", help="Used to force normal vram use if lowvram gets automatically enabled.")
+vram_group.add_argument("--lowvram", action="store_true", help="Split the unet in parts to use less vram.")
+vram_group.add_argument("--novram", action="store_true", help="When lowvram isn't enough.")
+vram_group.add_argument("--cpu", action="store_true", help="To use the CPU for everything (slow).")
+
+
+parser.add_argument("--disable-smart-memory", action="store_true", help="Force fcbh_backend to agressively offload to regular ram instead of keeping models in vram when it can.")
+
+
+parser.add_argument("--dont-print-server", action="store_true", help="Don't print server output.")
+parser.add_argument("--quick-test-for-ci", action="store_true", help="Quick test for CI.")
+parser.add_argument("--windows-standalone-build", action="store_true", help="Windows standalone build: Enable convenient things that most people using the standalone windows build will probably enjoy (like auto opening the page on startup).")
+
+parser.add_argument("--disable-metadata", action="store_true", help="Disable saving prompt metadata in files.")
+
+if fcbh.options.args_parsing:
+    args = parser.parse_args()
+else:
+    args = parser.parse_args([])
+
+if args.windows_standalone_build:
+    args.auto_launch = True
+
+if args.disable_auto_launch:
+    args.auto_launch = False

backend/headless/fcbh/clip_config_bigg.json

@@ -0,0 +1,23 @@
+{
+  "architectures": [
+    "CLIPTextModel"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 0,
+  "dropout": 0.0,
+  "eos_token_id": 2,
+  "hidden_act": "gelu",
+  "hidden_size": 1280,
+  "initializer_factor": 1.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 5120,
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 77,
+  "model_type": "clip_text_model",
+  "num_attention_heads": 20,
+  "num_hidden_layers": 32,
+  "pad_token_id": 1,
+  "projection_dim": 1280,
+  "torch_dtype": "float32",
+  "vocab_size": 49408
+}

backend/headless/fcbh/clip_vision.py

@@ -0,0 +1,116 @@
+from transformers import CLIPVisionModelWithProjection, CLIPVisionConfig, modeling_utils
+from .utils import load_torch_file, transformers_convert, common_upscale
+import os
+import torch
+import contextlib
+
+import fcbh.ops
+import fcbh.model_patcher
+import fcbh.model_management
+import fcbh.utils
+
+def clip_preprocess(image, size=224):
+    mean = torch.tensor([ 0.48145466,0.4578275,0.40821073], device=image.device, dtype=image.dtype)
+    std = torch.tensor([0.26862954,0.26130258,0.27577711], device=image.device, dtype=image.dtype)
+    scale = (size / min(image.shape[1], image.shape[2]))
+    image = torch.nn.functional.interpolate(image.movedim(-1, 1), size=(round(scale * image.shape[1]), round(scale * image.shape[2])), mode="bicubic", antialias=True)
+    h = (image.shape[2] - size)//2
+    w = (image.shape[3] - size)//2
+    image = image[:,:,h:h+size,w:w+size]
+    image = torch.clip((255. * image), 0, 255).round() / 255.0
+    return (image - mean.view([3,1,1])) / std.view([3,1,1])
+
+class ClipVisionModel():
+    def __init__(self, json_config):
+        config = CLIPVisionConfig.from_json_file(json_config)
+        self.load_device = fcbh.model_management.text_encoder_device()
+        offload_device = fcbh.model_management.text_encoder_offload_device()
+        self.dtype = torch.float32
+        if fcbh.model_management.should_use_fp16(self.load_device, prioritize_performance=False):
+            self.dtype = torch.float16
+
+        with fcbh.ops.use_fcbh_ops(offload_device, self.dtype):
+            with modeling_utils.no_init_weights():
+                self.model = CLIPVisionModelWithProjection(config)
+        self.model.to(self.dtype)
+
+        self.patcher = fcbh.model_patcher.ModelPatcher(self.model, load_device=self.load_device, offload_device=offload_device)
+    def load_sd(self, sd):
+        return self.model.load_state_dict(sd, strict=False)
+
+    def encode_image(self, image):
+        fcbh.model_management.load_model_gpu(self.patcher)
+        pixel_values = clip_preprocess(image.to(self.load_device))
+
+        if self.dtype != torch.float32:
+            precision_scope = torch.autocast
+        else:
+            precision_scope = lambda a, b: contextlib.nullcontext(a)
+
+        with precision_scope(fcbh.model_management.get_autocast_device(self.load_device), torch.float32):
+            outputs = self.model(pixel_values=pixel_values, output_hidden_states=True)
+
+        for k in outputs:
+            t = outputs[k]
+            if t is not None:
+                if k == 'hidden_states':
+                    outputs["penultimate_hidden_states"] = t[-2].cpu()
+                    outputs["hidden_states"] = None
+                else:
+                    outputs[k] = t.cpu()
+
+        return outputs
+
+def convert_to_transformers(sd, prefix):
+    sd_k = sd.keys()
+    if "{}transformer.resblocks.0.attn.in_proj_weight".format(prefix) in sd_k:
+        keys_to_replace = {
+            "{}class_embedding".format(prefix): "vision_model.embeddings.class_embedding",
+            "{}conv1.weight".format(prefix): "vision_model.embeddings.patch_embedding.weight",
+            "{}positional_embedding".format(prefix): "vision_model.embeddings.position_embedding.weight",
+            "{}ln_post.bias".format(prefix): "vision_model.post_layernorm.bias",
+            "{}ln_post.weight".format(prefix): "vision_model.post_layernorm.weight",
+            "{}ln_pre.bias".format(prefix): "vision_model.pre_layrnorm.bias",
+            "{}ln_pre.weight".format(prefix): "vision_model.pre_layrnorm.weight",
+        }
+
+        for x in keys_to_replace:
+            if x in sd_k:
+                sd[keys_to_replace[x]] = sd.pop(x)
+
+        if "{}proj".format(prefix) in sd_k:
+            sd['visual_projection.weight'] = sd.pop("{}proj".format(prefix)).transpose(0, 1)
+
+        sd = transformers_convert(sd, prefix, "vision_model.", 48)
+    return sd
+
+def load_clipvision_from_sd(sd, prefix="", convert_keys=False):
+    if convert_keys:
+        sd = convert_to_transformers(sd, prefix)
+    if "vision_model.encoder.layers.47.layer_norm1.weight" in sd:
+        json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "clip_vision_config_g.json")
+    elif "vision_model.encoder.layers.30.layer_norm1.weight" in sd:
+        json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "clip_vision_config_h.json")
+    elif "vision_model.encoder.layers.22.layer_norm1.weight" in sd:
+        json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "clip_vision_config_vitl.json")
+    else:
+        return None
+
+    clip = ClipVisionModel(json_config)
+    m, u = clip.load_sd(sd)
+    if len(m) > 0:
+        print("missing clip vision:", m)
+    u = set(u)
+    keys = list(sd.keys())
+    for k in keys:
+        if k not in u:
+            t = sd.pop(k)
+            del t
+    return clip
+
+def load(ckpt_path):
+    sd = load_torch_file(ckpt_path)
+    if "visual.transformer.resblocks.0.attn.in_proj_weight" in sd:
+        return load_clipvision_from_sd(sd, prefix="visual.", convert_keys=True)
+    else:
+        return load_clipvision_from_sd(sd)

backend/headless/fcbh/clip_vision_config_g.json

@@ -0,0 +1,18 @@
+{
+  "attention_dropout": 0.0,
+  "dropout": 0.0,
+  "hidden_act": "gelu",
+  "hidden_size": 1664,
+  "image_size": 224,
+  "initializer_factor": 1.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 8192,
+  "layer_norm_eps": 1e-05,
+  "model_type": "clip_vision_model",
+  "num_attention_heads": 16,
+  "num_channels": 3,
+  "num_hidden_layers": 48,
+  "patch_size": 14,
+  "projection_dim": 1280,
+  "torch_dtype": "float32"
+}

backend/headless/fcbh/clip_vision_config_h.json

@@ -0,0 +1,18 @@
+{
+  "attention_dropout": 0.0,
+  "dropout": 0.0,
+  "hidden_act": "gelu",
+  "hidden_size": 1280,
+  "image_size": 224,
+  "initializer_factor": 1.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 5120,
+  "layer_norm_eps": 1e-05,
+  "model_type": "clip_vision_model",
+  "num_attention_heads": 16,
+  "num_channels": 3,
+  "num_hidden_layers": 32,
+  "patch_size": 14,
+  "projection_dim": 1024,
+  "torch_dtype": "float32"
+}

backend/headless/fcbh/clip_vision_config_vitl.json

@@ -0,0 +1,18 @@
+{
+  "attention_dropout": 0.0,
+  "dropout": 0.0,
+  "hidden_act": "quick_gelu",
+  "hidden_size": 1024,
+  "image_size": 224,
+  "initializer_factor": 1.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "model_type": "clip_vision_model",
+  "num_attention_heads": 16,
+  "num_channels": 3,
+  "num_hidden_layers": 24,
+  "patch_size": 14,
+  "projection_dim": 768,
+  "torch_dtype": "float32"
+}

backend/headless/fcbh/conds.py

@@ -0,0 +1,79 @@
+import enum
+import torch
+import math
+import fcbh.utils
+
+
+def lcm(a, b): #TODO: eventually replace by math.lcm (added in python3.9)
+    return abs(a*b) // math.gcd(a, b)
+
+class CONDRegular:
+    def __init__(self, cond):
+        self.cond = cond
+
+    def _copy_with(self, cond):
+        return self.__class__(cond)
+
+    def process_cond(self, batch_size, device, **kwargs):
+        return self._copy_with(fcbh.utils.repeat_to_batch_size(self.cond, batch_size).to(device))
+
+    def can_concat(self, other):
+        if self.cond.shape != other.cond.shape:
+            return False
+        return True
+
+    def concat(self, others):
+        conds = [self.cond]
+        for x in others:
+            conds.append(x.cond)
+        return torch.cat(conds)
+
+class CONDNoiseShape(CONDRegular):
+    def process_cond(self, batch_size, device, area, **kwargs):
+        data = self.cond[:,:,area[2]:area[0] + area[2],area[3]:area[1] + area[3]]
+        return self._copy_with(fcbh.utils.repeat_to_batch_size(data, batch_size).to(device))
+
+
+class CONDCrossAttn(CONDRegular):
+    def can_concat(self, other):
+        s1 = self.cond.shape
+        s2 = other.cond.shape
+        if s1 != s2:
+            if s1[0] != s2[0] or s1[2] != s2[2]: #these 2 cases should not happen
+                return False
+
+            mult_min = lcm(s1[1], s2[1])
+            diff = mult_min // min(s1[1], s2[1])
+            if diff > 4: #arbitrary limit on the padding because it's probably going to impact performance negatively if it's too much
+                return False
+        return True
+
+    def concat(self, others):
+        conds = [self.cond]
+        crossattn_max_len = self.cond.shape[1]
+        for x in others:
+            c = x.cond
+            crossattn_max_len = lcm(crossattn_max_len, c.shape[1])
+            conds.append(c)
+
+        out = []
+        for c in conds:
+            if c.shape[1] < crossattn_max_len:
+                c = c.repeat(1, crossattn_max_len // c.shape[1], 1) #padding with repeat doesn't change result
+            out.append(c)
+        return torch.cat(out)
+
+class CONDConstant(CONDRegular):
+    def __init__(self, cond):
+        self.cond = cond
+
+    def process_cond(self, batch_size, device, **kwargs):
+        return self._copy_with(self.cond)
+
+    def can_concat(self, other):
+        if self.cond != other.cond:
+            return False
+        return True
+
+    def concat(self, others):
+        return self.cond

backend/headless/fcbh/controlnet.py

@@ -0,0 +1,499 @@
+import torch
+import math
+import os
+import fcbh.utils
+import fcbh.model_management
+import fcbh.model_detection
+import fcbh.model_patcher
+
+import fcbh.cldm.cldm
+import fcbh.t2i_adapter.adapter
+
+
+def broadcast_image_to(tensor, target_batch_size, batched_number):
+    current_batch_size = tensor.shape[0]
+    #print(current_batch_size, target_batch_size)
+    if current_batch_size == 1:
+        return tensor
+
+    per_batch = target_batch_size // batched_number
+    tensor = tensor[:per_batch]
+
+    if per_batch > tensor.shape[0]:
+        tensor = torch.cat([tensor] * (per_batch // tensor.shape[0]) + [tensor[:(per_batch % tensor.shape[0])]], dim=0)
+
+    current_batch_size = tensor.shape[0]
+    if current_batch_size == target_batch_size:
+        return tensor
+    else:
+        return torch.cat([tensor] * batched_number, dim=0)
+
+class ControlBase:
+    def __init__(self, device=None):
+        self.cond_hint_original = None
+        self.cond_hint = None
+        self.strength = 1.0
+        self.timestep_percent_range = (1.0, 0.0)
+        self.timestep_range = None
+
+        if device is None:
+            device = fcbh.model_management.get_torch_device()
+        self.device = device
+        self.previous_controlnet = None
+        self.global_average_pooling = False
+
+    def set_cond_hint(self, cond_hint, strength=1.0, timestep_percent_range=(1.0, 0.0)):
+        self.cond_hint_original = cond_hint
+        self.strength = strength
+        self.timestep_percent_range = timestep_percent_range
+        return self
+
+    def pre_run(self, model, percent_to_timestep_function):
+        self.timestep_range = (percent_to_timestep_function(self.timestep_percent_range[0]), percent_to_timestep_function(self.timestep_percent_range[1]))
+        if self.previous_controlnet is not None:
+            self.previous_controlnet.pre_run(model, percent_to_timestep_function)
+
+    def set_previous_controlnet(self, controlnet):
+        self.previous_controlnet = controlnet
+        return self
+
+    def cleanup(self):
+        if self.previous_controlnet is not None:
+            self.previous_controlnet.cleanup()
+        if self.cond_hint is not None:
+            del self.cond_hint
+            self.cond_hint = None
+        self.timestep_range = None
+
+    def get_models(self):
+        out = []
+        if self.previous_controlnet is not None:
+            out += self.previous_controlnet.get_models()
+        return out
+
+    def copy_to(self, c):
+        c.cond_hint_original = self.cond_hint_original
+        c.strength = self.strength
+        c.timestep_percent_range = self.timestep_percent_range
+
+    def inference_memory_requirements(self, dtype):
+        if self.previous_controlnet is not None:
+            return self.previous_controlnet.inference_memory_requirements(dtype)
+        return 0
+
+    def control_merge(self, control_input, control_output, control_prev, output_dtype):
+        out = {'input':[], 'middle':[], 'output': []}
+
+        if control_input is not None:
+            for i in range(len(control_input)):
+                key = 'input'
+                x = control_input[i]
+                if x is not None:
+                    x *= self.strength
+                    if x.dtype != output_dtype:
+                        x = x.to(output_dtype)
+                out[key].insert(0, x)
+
+        if control_output is not None:
+            for i in range(len(control_output)):
+                if i == (len(control_output) - 1):
+                    key = 'middle'
+                    index = 0
+                else:
+                    key = 'output'
+                    index = i
+                x = control_output[i]
+                if x is not None:
+                    if self.global_average_pooling:
+                        x = torch.mean(x, dim=(2, 3), keepdim=True).repeat(1, 1, x.shape[2], x.shape[3])
+
+                    x *= self.strength
+                    if x.dtype != output_dtype:
+                        x = x.to(output_dtype)
+
+                out[key].append(x)
+        if control_prev is not None:
+            for x in ['input', 'middle', 'output']:
+                o = out[x]
+                for i in range(len(control_prev[x])):
+                    prev_val = control_prev[x][i]
+                    if i >= len(o):
+                        o.append(prev_val)
+                    elif prev_val is not None:
+                        if o[i] is None:
+                            o[i] = prev_val
+                        else:
+                            o[i] += prev_val
+        return out
+
+class ControlNet(ControlBase):
+    def __init__(self, control_model, global_average_pooling=False, device=None):
+        super().__init__(device)
+        self.control_model = control_model
+        self.control_model_wrapped = fcbh.model_patcher.ModelPatcher(self.control_model, load_device=fcbh.model_management.get_torch_device(), offload_device=fcbh.model_management.unet_offload_device())
+        self.global_average_pooling = global_average_pooling
+        self.model_sampling_current = None
+
+    def get_control(self, x_noisy, t, cond, batched_number):
+        control_prev = None
+        if self.previous_controlnet is not None:
+            control_prev = self.previous_controlnet.get_control(x_noisy, t, cond, batched_number)
+
+        if self.timestep_range is not None:
+            if t[0] > self.timestep_range[0] or t[0] < self.timestep_range[1]:
+                if control_prev is not None:
+                    return control_prev
+                else:
+                    return None
+
+        output_dtype = x_noisy.dtype
+        if self.cond_hint is None or x_noisy.shape[2] * 8 != self.cond_hint.shape[2] or x_noisy.shape[3] * 8 != self.cond_hint.shape[3]:
+            if self.cond_hint is not None:
+                del self.cond_hint
+            self.cond_hint = None
+            self.cond_hint = fcbh.utils.common_upscale(self.cond_hint_original, x_noisy.shape[3] * 8, x_noisy.shape[2] * 8, 'nearest-exact', "center").to(self.control_model.dtype).to(self.device)
+        if x_noisy.shape[0] != self.cond_hint.shape[0]:
+            self.cond_hint = broadcast_image_to(self.cond_hint, x_noisy.shape[0], batched_number)
+
+
+        context = cond['c_crossattn']
+        y = cond.get('y', None)
+        if y is not None:
+            y = y.to(self.control_model.dtype)
+        timestep = self.model_sampling_current.timestep(t)
+        x_noisy = self.model_sampling_current.calculate_input(t, x_noisy)
+
+        control = self.control_model(x=x_noisy.to(self.control_model.dtype), hint=self.cond_hint, timesteps=timestep.float(), context=context.to(self.control_model.dtype), y=y)
+        return self.control_merge(None, control, control_prev, output_dtype)
+
+    def copy(self):
+        c = ControlNet(self.control_model, global_average_pooling=self.global_average_pooling)
+        self.copy_to(c)
+        return c
+
+    def get_models(self):
+        out = super().get_models()
+        out.append(self.control_model_wrapped)
+        return out
+
+    def pre_run(self, model, percent_to_timestep_function):
+        super().pre_run(model, percent_to_timestep_function)
+        self.model_sampling_current = model.model_sampling
+
+    def cleanup(self):
+        self.model_sampling_current = None
+        super().cleanup()
+
+class ControlLoraOps:
+    class Linear(torch.nn.Module):
+        def __init__(self, in_features: int, out_features: int, bias: bool = True,
+                    device=None, dtype=None) -> None:
+            factory_kwargs = {'device': device, 'dtype': dtype}
+            super().__init__()
+            self.in_features = in_features
+            self.out_features = out_features
+            self.weight = None
+            self.up = None
+            self.down = None
+            self.bias = None
+
+        def forward(self, input):
+            if self.up is not None:
+                return torch.nn.functional.linear(input, self.weight.to(input.device) + (torch.mm(self.up.flatten(start_dim=1), self.down.flatten(start_dim=1))).reshape(self.weight.shape).type(input.dtype), self.bias)
+            else:
+                return torch.nn.functional.linear(input, self.weight.to(input.device), self.bias)
+
+    class Conv2d(torch.nn.Module):
+        def __init__(
+            self,
+            in_channels,
+            out_channels,
+            kernel_size,
+            stride=1,
+            padding=0,
+            dilation=1,
+            groups=1,
+            bias=True,
+            padding_mode='zeros',
+            device=None,
+            dtype=None
+        ):
+            super().__init__()
+            self.in_channels = in_channels
+            self.out_channels = out_channels
+            self.kernel_size = kernel_size
+            self.stride = stride
+            self.padding = padding
+            self.dilation = dilation
+            self.transposed = False
+            self.output_padding = 0
+            self.groups = groups
+            self.padding_mode = padding_mode
+
+            self.weight = None
+            self.bias = None
+            self.up = None
+            self.down = None
+
+
+        def forward(self, input):
+            if self.up is not None:
+                return torch.nn.functional.conv2d(input, self.weight.to(input.device) + (torch.mm(self.up.flatten(start_dim=1), self.down.flatten(start_dim=1))).reshape(self.weight.shape).type(input.dtype), self.bias, self.stride, self.padding, self.dilation, self.groups)
+            else:
+                return torch.nn.functional.conv2d(input, self.weight.to(input.device), self.bias, self.stride, self.padding, self.dilation, self.groups)
+
+    def conv_nd(self, dims, *args, **kwargs):
+        if dims == 2:
+            return self.Conv2d(*args, **kwargs)
+        else:
+            raise ValueError(f"unsupported dimensions: {dims}")
+
+
+class ControlLora(ControlNet):
+    def __init__(self, control_weights, global_average_pooling=False, device=None):
+        ControlBase.__init__(self, device)
+        self.control_weights = control_weights
+        self.global_average_pooling = global_average_pooling
+
+    def pre_run(self, model, percent_to_timestep_function):
+        super().pre_run(model, percent_to_timestep_function)
+        controlnet_config = model.model_config.unet_config.copy()
+        controlnet_config.pop("out_channels")
+        controlnet_config["hint_channels"] = self.control_weights["input_hint_block.0.weight"].shape[1]
+        controlnet_config["operations"] = ControlLoraOps()
+        self.control_model = fcbh.cldm.cldm.ControlNet(**controlnet_config)
+        dtype = model.get_dtype()
+        self.control_model.to(dtype)
+        self.control_model.to(fcbh.model_management.get_torch_device())
+        diffusion_model = model.diffusion_model
+        sd = diffusion_model.state_dict()
+        cm = self.control_model.state_dict()
+
+        for k in sd:
+            weight = fcbh.model_management.resolve_lowvram_weight(sd[k], diffusion_model, k)
+            try:
+                fcbh.utils.set_attr(self.control_model, k, weight)
+            except:
+                pass
+
+        for k in self.control_weights:
+            if k not in {"lora_controlnet"}:
+                fcbh.utils.set_attr(self.control_model, k, self.control_weights[k].to(dtype).to(fcbh.model_management.get_torch_device()))
+
+    def copy(self):
+        c = ControlLora(self.control_weights, global_average_pooling=self.global_average_pooling)
+        self.copy_to(c)
+        return c
+
+    def cleanup(self):
+        del self.control_model
+        self.control_model = None
+        super().cleanup()
+
+    def get_models(self):
+        out = ControlBase.get_models(self)
+        return out
+
+    def inference_memory_requirements(self, dtype):
+        return fcbh.utils.calculate_parameters(self.control_weights) * fcbh.model_management.dtype_size(dtype) + ControlBase.inference_memory_requirements(self, dtype)
+
+def load_controlnet(ckpt_path, model=None):
+    controlnet_data = fcbh.utils.load_torch_file(ckpt_path, safe_load=True)
+    if "lora_controlnet" in controlnet_data:
+        return ControlLora(controlnet_data)
+
+    controlnet_config = None
+    if "controlnet_cond_embedding.conv_in.weight" in controlnet_data: #diffusers format
+        unet_dtype = fcbh.model_management.unet_dtype()
+        controlnet_config = fcbh.model_detection.unet_config_from_diffusers_unet(controlnet_data, unet_dtype)
+        diffusers_keys = fcbh.utils.unet_to_diffusers(controlnet_config)
+        diffusers_keys["controlnet_mid_block.weight"] = "middle_block_out.0.weight"
+        diffusers_keys["controlnet_mid_block.bias"] = "middle_block_out.0.bias"
+
+        count = 0
+        loop = True
+        while loop:
+            suffix = [".weight", ".bias"]
+            for s in suffix:
+                k_in = "controlnet_down_blocks.{}{}".format(count, s)
+                k_out = "zero_convs.{}.0{}".format(count, s)
+                if k_in not in controlnet_data:
+                    loop = False
+                    break
+                diffusers_keys[k_in] = k_out
+            count += 1
+
+        count = 0
+        loop = True
+        while loop:
+            suffix = [".weight", ".bias"]
+            for s in suffix:
+                if count == 0:
+                    k_in = "controlnet_cond_embedding.conv_in{}".format(s)
+                else:
+                    k_in = "controlnet_cond_embedding.blocks.{}{}".format(count - 1, s)
+                k_out = "input_hint_block.{}{}".format(count * 2, s)
+                if k_in not in controlnet_data:
+                    k_in = "controlnet_cond_embedding.conv_out{}".format(s)
+                    loop = False
+                diffusers_keys[k_in] = k_out
+            count += 1
+
+        new_sd = {}
+        for k in diffusers_keys:
+            if k in controlnet_data:
+                new_sd[diffusers_keys[k]] = controlnet_data.pop(k)
+
+        leftover_keys = controlnet_data.keys()
+        if len(leftover_keys) > 0:
+            print("leftover keys:", leftover_keys)
+        controlnet_data = new_sd
+
+    pth_key = 'control_model.zero_convs.0.0.weight'
+    pth = False
+    key = 'zero_convs.0.0.weight'
+    if pth_key in controlnet_data:
+        pth = True
+        key = pth_key
+        prefix = "control_model."
+    elif key in controlnet_data:
+        prefix = ""
+    else:
+        net = load_t2i_adapter(controlnet_data)
+        if net is None:
+            print("error checkpoint does not contain controlnet or t2i adapter data", ckpt_path)
+        return net
+
+    if controlnet_config is None:
+        unet_dtype = fcbh.model_management.unet_dtype()
+        controlnet_config = fcbh.model_detection.model_config_from_unet(controlnet_data, prefix, unet_dtype, True).unet_config
+    controlnet_config.pop("out_channels")
+    controlnet_config["hint_channels"] = controlnet_data["{}input_hint_block.0.weight".format(prefix)].shape[1]
+    control_model = fcbh.cldm.cldm.ControlNet(**controlnet_config)
+
+    if pth:
+        if 'difference' in controlnet_data:
+            if model is not None:
+                fcbh.model_management.load_models_gpu([model])
+                model_sd = model.model_state_dict()
+                for x in controlnet_data:
+                    c_m = "control_model."
+                    if x.startswith(c_m):
+                        sd_key = "diffusion_model.{}".format(x[len(c_m):])
+                        if sd_key in model_sd:
+                            cd = controlnet_data[x]
+                            cd += model_sd[sd_key].type(cd.dtype).to(cd.device)
+            else:
+                print("WARNING: Loaded a diff controlnet without a model. It will very likely not work.")
+
+        class WeightsLoader(torch.nn.Module):
+            pass
+        w = WeightsLoader()
+        w.control_model = control_model
+        missing, unexpected = w.load_state_dict(controlnet_data, strict=False)
+    else:
+        missing, unexpected = control_model.load_state_dict(controlnet_data, strict=False)
+    print(missing, unexpected)
+
+    control_model = control_model.to(unet_dtype)
+
+    global_average_pooling = False
+    filename = os.path.splitext(ckpt_path)[0]
+    if filename.endswith("_shuffle") or filename.endswith("_shuffle_fp16"): #TODO: smarter way of enabling global_average_pooling
+        global_average_pooling = True
+
+    control = ControlNet(control_model, global_average_pooling=global_average_pooling)
+    return control
+
+class T2IAdapter(ControlBase):
+    def __init__(self, t2i_model, channels_in, device=None):
+        super().__init__(device)
+        self.t2i_model = t2i_model
+        self.channels_in = channels_in
+        self.control_input = None
+
+    def scale_image_to(self, width, height):
+        unshuffle_amount = self.t2i_model.unshuffle_amount
+        width = math.ceil(width / unshuffle_amount) * unshuffle_amount
+        height = math.ceil(height / unshuffle_amount) * unshuffle_amount
+        return width, height
+
+    def get_control(self, x_noisy, t, cond, batched_number):
+        control_prev = None
+        if self.previous_controlnet is not None:
+            control_prev = self.previous_controlnet.get_control(x_noisy, t, cond, batched_number)
+
+        if self.timestep_range is not None:
+            if t[0] > self.timestep_range[0] or t[0] < self.timestep_range[1]:
+                if control_prev is not None:
+                    return control_prev
+                else:
+                    return None
+
+        if self.cond_hint is None or x_noisy.shape[2] * 8 != self.cond_hint.shape[2] or x_noisy.shape[3] * 8 != self.cond_hint.shape[3]:
+            if self.cond_hint is not None:
+                del self.cond_hint
+            self.control_input = None
+            self.cond_hint = None
+            width, height = self.scale_image_to(x_noisy.shape[3] * 8, x_noisy.shape[2] * 8)
+            self.cond_hint = fcbh.utils.common_upscale(self.cond_hint_original, width, height, 'nearest-exact', "center").float().to(self.device)
+            if self.channels_in == 1 and self.cond_hint.shape[1] > 1:
+                self.cond_hint = torch.mean(self.cond_hint, 1, keepdim=True)
+        if x_noisy.shape[0] != self.cond_hint.shape[0]:
+            self.cond_hint = broadcast_image_to(self.cond_hint, x_noisy.shape[0], batched_number)
+        if self.control_input is None:
+            self.t2i_model.to(x_noisy.dtype)
+            self.t2i_model.to(self.device)
+            self.control_input = self.t2i_model(self.cond_hint.to(x_noisy.dtype))
+            self.t2i_model.cpu()
+
+        control_input = list(map(lambda a: None if a is None else a.clone(), self.control_input))
+        mid = None
+        if self.t2i_model.xl == True:
+            mid = control_input[-1:]
+            control_input = control_input[:-1]
+        return self.control_merge(control_input, mid, control_prev, x_noisy.dtype)
+
+    def copy(self):
+        c = T2IAdapter(self.t2i_model, self.channels_in)
+        self.copy_to(c)
+        return c
+
+def load_t2i_adapter(t2i_data):
+    if 'adapter' in t2i_data:
+        t2i_data = t2i_data['adapter']
+    if 'adapter.body.0.resnets.0.block1.weight' in t2i_data: #diffusers format
+        prefix_replace = {}
+        for i in range(4):
+            for j in range(2):
+                prefix_replace["adapter.body.{}.resnets.{}.".format(i, j)] = "body.{}.".format(i * 2 + j)
+            prefix_replace["adapter.body.{}.".format(i, j)] = "body.{}.".format(i * 2)
+        prefix_replace["adapter."] = ""
+        t2i_data = fcbh.utils.state_dict_prefix_replace(t2i_data, prefix_replace)
+    keys = t2i_data.keys()
+
+    if "body.0.in_conv.weight" in keys:
+        cin = t2i_data['body.0.in_conv.weight'].shape[1]
+        model_ad = fcbh.t2i_adapter.adapter.Adapter_light(cin=cin, channels=[320, 640, 1280, 1280], nums_rb=4)
+    elif 'conv_in.weight' in keys:
+        cin = t2i_data['conv_in.weight'].shape[1]
+        channel = t2i_data['conv_in.weight'].shape[0]
+        ksize = t2i_data['body.0.block2.weight'].shape[2]
+        use_conv = False
+        down_opts = list(filter(lambda a: a.endswith("down_opt.op.weight"), keys))
+        if len(down_opts) > 0:
+            use_conv = True
+        xl = False
+        if cin == 256 or cin == 768:
+            xl = True
+        model_ad = fcbh.t2i_adapter.adapter.Adapter(cin=cin, channels=[channel, channel*2, channel*4, channel*4][:4], nums_rb=2, ksize=ksize, sk=True, use_conv=use_conv, xl=xl)
+    else:
+        return None
+    missing, unexpected = model_ad.load_state_dict(t2i_data)
+    if len(missing) > 0:
+        print("t2i missing", missing)
+
+    if len(unexpected) > 0:
+        print("t2i unexpected", unexpected)
+
+    return T2IAdapter(model_ad, model_ad.input_channels)

backend/headless/fcbh/diffusers_convert.py

@@ -0,0 +1,261 @@
+import re
+import torch
+
+# conversion code from https://github.com/huggingface/diffusers/blob/main/scripts/convert_diffusers_to_original_stable_diffusion.py
+
+# =================#
+# UNet Conversion #
+# =================#
+
+unet_conversion_map = [
+    # (stable-diffusion, HF Diffusers)
+    ("time_embed.0.weight", "time_embedding.linear_1.weight"),
+    ("time_embed.0.bias", "time_embedding.linear_1.bias"),
+    ("time_embed.2.weight", "time_embedding.linear_2.weight"),
+    ("time_embed.2.bias", "time_embedding.linear_2.bias"),
+    ("input_blocks.0.0.weight", "conv_in.weight"),
+    ("input_blocks.0.0.bias", "conv_in.bias"),
+    ("out.0.weight", "conv_norm_out.weight"),
+    ("out.0.bias", "conv_norm_out.bias"),
+    ("out.2.weight", "conv_out.weight"),
+    ("out.2.bias", "conv_out.bias"),
+]
+
+unet_conversion_map_resnet = [
+    # (stable-diffusion, HF Diffusers)
+    ("in_layers.0", "norm1"),
+    ("in_layers.2", "conv1"),
+    ("out_layers.0", "norm2"),
+    ("out_layers.3", "conv2"),
+    ("emb_layers.1", "time_emb_proj"),
+    ("skip_connection", "conv_shortcut"),
+]
+
+unet_conversion_map_layer = []
+# hardcoded number of downblocks and resnets/attentions...
+# would need smarter logic for other networks.
+for i in range(4):
+    # loop over downblocks/upblocks
+
+    for j in range(2):
+        # loop over resnets/attentions for downblocks
+        hf_down_res_prefix = f"down_blocks.{i}.resnets.{j}."
+        sd_down_res_prefix = f"input_blocks.{3 * i + j + 1}.0."
+        unet_conversion_map_layer.append((sd_down_res_prefix, hf_down_res_prefix))
+
+        if i < 3:
+            # no attention layers in down_blocks.3
+            hf_down_atn_prefix = f"down_blocks.{i}.attentions.{j}."
+            sd_down_atn_prefix = f"input_blocks.{3 * i + j + 1}.1."
+            unet_conversion_map_layer.append((sd_down_atn_prefix, hf_down_atn_prefix))
+
+    for j in range(3):
+        # loop over resnets/attentions for upblocks
+        hf_up_res_prefix = f"up_blocks.{i}.resnets.{j}."
+        sd_up_res_prefix = f"output_blocks.{3 * i + j}.0."
+        unet_conversion_map_layer.append((sd_up_res_prefix, hf_up_res_prefix))
+
+        if i > 0:
+            # no attention layers in up_blocks.0
+            hf_up_atn_prefix = f"up_blocks.{i}.attentions.{j}."
+            sd_up_atn_prefix = f"output_blocks.{3 * i + j}.1."
+            unet_conversion_map_layer.append((sd_up_atn_prefix, hf_up_atn_prefix))
+
+    if i < 3:
+        # no downsample in down_blocks.3
+        hf_downsample_prefix = f"down_blocks.{i}.downsamplers.0.conv."
+        sd_downsample_prefix = f"input_blocks.{3 * (i + 1)}.0.op."
+        unet_conversion_map_layer.append((sd_downsample_prefix, hf_downsample_prefix))
+
+        # no upsample in up_blocks.3
+        hf_upsample_prefix = f"up_blocks.{i}.upsamplers.0."
+        sd_upsample_prefix = f"output_blocks.{3 * i + 2}.{1 if i == 0 else 2}."
+        unet_conversion_map_layer.append((sd_upsample_prefix, hf_upsample_prefix))
+
+hf_mid_atn_prefix = "mid_block.attentions.0."
+sd_mid_atn_prefix = "middle_block.1."
+unet_conversion_map_layer.append((sd_mid_atn_prefix, hf_mid_atn_prefix))
+
+for j in range(2):
+    hf_mid_res_prefix = f"mid_block.resnets.{j}."
+    sd_mid_res_prefix = f"middle_block.{2 * j}."
+    unet_conversion_map_layer.append((sd_mid_res_prefix, hf_mid_res_prefix))
+
+
+def convert_unet_state_dict(unet_state_dict):
+    # buyer beware: this is a *brittle* function,
+    # and correct output requires that all of these pieces interact in
+    # the exact order in which I have arranged them.
+    mapping = {k: k for k in unet_state_dict.keys()}
+    for sd_name, hf_name in unet_conversion_map:
+        mapping[hf_name] = sd_name
+    for k, v in mapping.items():
+        if "resnets" in k:
+            for sd_part, hf_part in unet_conversion_map_resnet:
+                v = v.replace(hf_part, sd_part)
+            mapping[k] = v
+    for k, v in mapping.items():
+        for sd_part, hf_part in unet_conversion_map_layer:
+            v = v.replace(hf_part, sd_part)
+        mapping[k] = v
+    new_state_dict = {v: unet_state_dict[k] for k, v in mapping.items()}
+    return new_state_dict
+
+
+# ================#
+# VAE Conversion #
+# ================#
+
+vae_conversion_map = [
+    # (stable-diffusion, HF Diffusers)
+    ("nin_shortcut", "conv_shortcut"),
+    ("norm_out", "conv_norm_out"),
+    ("mid.attn_1.", "mid_block.attentions.0."),
+]
+
+for i in range(4):
+    # down_blocks have two resnets
+    for j in range(2):
+        hf_down_prefix = f"encoder.down_blocks.{i}.resnets.{j}."
+        sd_down_prefix = f"encoder.down.{i}.block.{j}."
+        vae_conversion_map.append((sd_down_prefix, hf_down_prefix))
+
+    if i < 3:
+        hf_downsample_prefix = f"down_blocks.{i}.downsamplers.0."
+        sd_downsample_prefix = f"down.{i}.downsample."
+        vae_conversion_map.append((sd_downsample_prefix, hf_downsample_prefix))
+
+        hf_upsample_prefix = f"up_blocks.{i}.upsamplers.0."
+        sd_upsample_prefix = f"up.{3 - i}.upsample."
+        vae_conversion_map.append((sd_upsample_prefix, hf_upsample_prefix))
+
+    # up_blocks have three resnets
+    # also, up blocks in hf are numbered in reverse from sd
+    for j in range(3):
+        hf_up_prefix = f"decoder.up_blocks.{i}.resnets.{j}."
+        sd_up_prefix = f"decoder.up.{3 - i}.block.{j}."
+        vae_conversion_map.append((sd_up_prefix, hf_up_prefix))
+
+# this part accounts for mid blocks in both the encoder and the decoder
+for i in range(2):
+    hf_mid_res_prefix = f"mid_block.resnets.{i}."
+    sd_mid_res_prefix = f"mid.block_{i + 1}."
+    vae_conversion_map.append((sd_mid_res_prefix, hf_mid_res_prefix))
+
+vae_conversion_map_attn = [
+    # (stable-diffusion, HF Diffusers)
+    ("norm.", "group_norm."),
+    ("q.", "query."),
+    ("k.", "key."),
+    ("v.", "value."),
+    ("q.", "to_q."),
+    ("k.", "to_k."),
+    ("v.", "to_v."),
+    ("proj_out.", "to_out.0."),
+    ("proj_out.", "proj_attn."),
+]
+
+
+def reshape_weight_for_sd(w):
+    # convert HF linear weights to SD conv2d weights
+    return w.reshape(*w.shape, 1, 1)
+
+
+def convert_vae_state_dict(vae_state_dict):
+    mapping = {k: k for k in vae_state_dict.keys()}
+    for k, v in mapping.items():
+        for sd_part, hf_part in vae_conversion_map:
+            v = v.replace(hf_part, sd_part)
+        mapping[k] = v
+    for k, v in mapping.items():
+        if "attentions" in k:
+            for sd_part, hf_part in vae_conversion_map_attn:
+                v = v.replace(hf_part, sd_part)
+            mapping[k] = v
+    new_state_dict = {v: vae_state_dict[k] for k, v in mapping.items()}
+    weights_to_convert = ["q", "k", "v", "proj_out"]
+    for k, v in new_state_dict.items():
+        for weight_name in weights_to_convert:
+            if f"mid.attn_1.{weight_name}.weight" in k:
+                print(f"Reshaping {k} for SD format")
+                new_state_dict[k] = reshape_weight_for_sd(v)
+    return new_state_dict
+
+
+# =========================#
+# Text Encoder Conversion #
+# =========================#
+
+
+textenc_conversion_lst = [
+    # (stable-diffusion, HF Diffusers)
+    ("resblocks.", "text_model.encoder.layers."),
+    ("ln_1", "layer_norm1"),
+    ("ln_2", "layer_norm2"),
+    (".c_fc.", ".fc1."),
+    (".c_proj.", ".fc2."),
+    (".attn", ".self_attn"),
+    ("ln_final.", "transformer.text_model.final_layer_norm."),
+    ("token_embedding.weight", "transformer.text_model.embeddings.token_embedding.weight"),
+    ("positional_embedding", "transformer.text_model.embeddings.position_embedding.weight"),
+]
+protected = {re.escape(x[1]): x[0] for x in textenc_conversion_lst}
+textenc_pattern = re.compile("|".join(protected.keys()))
+
+# Ordering is from https://github.com/pytorch/pytorch/blob/master/test/cpp/api/modules.cpp
+code2idx = {"q": 0, "k": 1, "v": 2}
+
+
+def convert_text_enc_state_dict_v20(text_enc_dict, prefix=""):
+    new_state_dict = {}
+    capture_qkv_weight = {}
+    capture_qkv_bias = {}
+    for k, v in text_enc_dict.items():
+        if not k.startswith(prefix):
+            continue
+        if (
+                k.endswith(".self_attn.q_proj.weight")
+                or k.endswith(".self_attn.k_proj.weight")
+                or k.endswith(".self_attn.v_proj.weight")
+        ):
+            k_pre = k[: -len(".q_proj.weight")]
+            k_code = k[-len("q_proj.weight")]
+            if k_pre not in capture_qkv_weight:
+                capture_qkv_weight[k_pre] = [None, None, None]
+            capture_qkv_weight[k_pre][code2idx[k_code]] = v
+            continue
+
+        if (
+                k.endswith(".self_attn.q_proj.bias")
+                or k.endswith(".self_attn.k_proj.bias")
+                or k.endswith(".self_attn.v_proj.bias")
+        ):
+            k_pre = k[: -len(".q_proj.bias")]
+            k_code = k[-len("q_proj.bias")]
+            if k_pre not in capture_qkv_bias:
+                capture_qkv_bias[k_pre] = [None, None, None]
+            capture_qkv_bias[k_pre][code2idx[k_code]] = v
+            continue
+
+        relabelled_key = textenc_pattern.sub(lambda m: protected[re.escape(m.group(0))], k)
+        new_state_dict[relabelled_key] = v
+
+    for k_pre, tensors in capture_qkv_weight.items():
+        if None in tensors:
+            raise Exception("CORRUPTED MODEL: one of the q-k-v values for the text encoder was missing")
+        relabelled_key = textenc_pattern.sub(lambda m: protected[re.escape(m.group(0))], k_pre)
+        new_state_dict[relabelled_key + ".in_proj_weight"] = torch.cat(tensors)
+
+    for k_pre, tensors in capture_qkv_bias.items():
+        if None in tensors:
+            raise Exception("CORRUPTED MODEL: one of the q-k-v values for the text encoder was missing")
+        relabelled_key = textenc_pattern.sub(lambda m: protected[re.escape(m.group(0))], k_pre)
+        new_state_dict[relabelled_key + ".in_proj_bias"] = torch.cat(tensors)
+
+    return new_state_dict
+
+
+def convert_text_enc_state_dict(text_enc_dict):
+    return text_enc_dict
+
+

backend/headless/fcbh/diffusers_load.py

@@ -0,0 +1,37 @@
+import json
+import os
+
+import fcbh.sd
+
+def first_file(path, filenames):
+    for f in filenames:
+        p = os.path.join(path, f)
+        if os.path.exists(p):
+            return p
+    return None
+
+def load_diffusers(model_path, output_vae=True, output_clip=True, embedding_directory=None):
+    diffusion_model_names = ["diffusion_pytorch_model.fp16.safetensors", "diffusion_pytorch_model.safetensors", "diffusion_pytorch_model.fp16.bin", "diffusion_pytorch_model.bin"]
+    unet_path = first_file(os.path.join(model_path, "unet"), diffusion_model_names)
+    vae_path = first_file(os.path.join(model_path, "vae"), diffusion_model_names)
+
+    text_encoder_model_names = ["model.fp16.safetensors", "model.safetensors", "pytorch_model.fp16.bin", "pytorch_model.bin"]
+    text_encoder1_path = first_file(os.path.join(model_path, "text_encoder"), text_encoder_model_names)
+    text_encoder2_path = first_file(os.path.join(model_path, "text_encoder_2"), text_encoder_model_names)
+
+    text_encoder_paths = [text_encoder1_path]
+    if text_encoder2_path is not None:
+        text_encoder_paths.append(text_encoder2_path)
+
+    unet = fcbh.sd.load_unet(unet_path)
+
+    clip = None
+    if output_clip:
+        clip = fcbh.sd.load_clip(text_encoder_paths, embedding_directory=embedding_directory)
+
+    vae = None
+    if output_vae:
+        sd = fcbh.utils.load_torch_file(vae_path)
+        vae = fcbh.sd.VAE(sd=sd)
+
+    return (unet, clip, vae)

backend/headless/fcbh/extra_samplers/uni_pc.py

@@ -0,0 +1,894 @@
+#code taken from: https://github.com/wl-zhao/UniPC and modified
+
+import torch
+import torch.nn.functional as F
+import math
+
+from tqdm.auto import trange, tqdm
+
+
+class NoiseScheduleVP:
+    def __init__(
+            self,
+            schedule='discrete',
+            betas=None,
+            alphas_cumprod=None,
+            continuous_beta_0=0.1,
+            continuous_beta_1=20.,
+        ):
+        """Create a wrapper class for the forward SDE (VP type).
+
+        ***
+        Update: We support discrete-time diffusion models by implementing a picewise linear interpolation for log_alpha_t.
+                We recommend to use schedule='discrete' for the discrete-time diffusion models, especially for high-resolution images.
+        ***
+
+        The forward SDE ensures that the condition distribution q_{t|0}(x_t | x_0) = N ( alpha_t * x_0, sigma_t^2 * I ).
+        We further define lambda_t = log(alpha_t) - log(sigma_t), which is the half-logSNR (described in the DPM-Solver paper).
+        Therefore, we implement the functions for computing alpha_t, sigma_t and lambda_t. For t in [0, T], we have:
+
+            log_alpha_t = self.marginal_log_mean_coeff(t)
+            sigma_t = self.marginal_std(t)
+            lambda_t = self.marginal_lambda(t)
+
+        Moreover, as lambda(t) is an invertible function, we also support its inverse function:
+
+            t = self.inverse_lambda(lambda_t)
+
+        ===============================================================
+
+        We support both discrete-time DPMs (trained on n = 0, 1, ..., N-1) and continuous-time DPMs (trained on t in [t_0, T]).
+
+        1. For discrete-time DPMs:
+
+            For discrete-time DPMs trained on n = 0, 1, ..., N-1, we convert the discrete steps to continuous time steps by:
+                t_i = (i + 1) / N
+            e.g. for N = 1000, we have t_0 = 1e-3 and T = t_{N-1} = 1.
+            We solve the corresponding diffusion ODE from time T = 1 to time t_0 = 1e-3.
+
+            Args:
+                betas: A `torch.Tensor`. The beta array for the discrete-time DPM. (See the original DDPM paper for details)
+                alphas_cumprod: A `torch.Tensor`. The cumprod alphas for the discrete-time DPM. (See the original DDPM paper for details)
+
+            Note that we always have alphas_cumprod = cumprod(betas). Therefore, we only need to set one of `betas` and `alphas_cumprod`.
+
+            **Important**:  Please pay special attention for the args for `alphas_cumprod`:
+                The `alphas_cumprod` is the \hat{alpha_n} arrays in the notations of DDPM. Specifically, DDPMs assume that
+                    q_{t_n | 0}(x_{t_n} | x_0) = N ( \sqrt{\hat{alpha_n}} * x_0, (1 - \hat{alpha_n}) * I ).
+                Therefore, the notation \hat{alpha_n} is different from the notation alpha_t in DPM-Solver. In fact, we have
+                    alpha_{t_n} = \sqrt{\hat{alpha_n}},
+                and
+                    log(alpha_{t_n}) = 0.5 * log(\hat{alpha_n}).
+
+
+        2. For continuous-time DPMs:
+
+            We support two types of VPSDEs: linear (DDPM) and cosine (improved-DDPM). The hyperparameters for the noise
+            schedule are the default settings in DDPM and improved-DDPM:
+
+            Args:
+                beta_min: A `float` number. The smallest beta for the linear schedule.
+                beta_max: A `float` number. The largest beta for the linear schedule.
+                cosine_s: A `float` number. The hyperparameter in the cosine schedule.
+                cosine_beta_max: A `float` number. The hyperparameter in the cosine schedule.
+                T: A `float` number. The ending time of the forward process.
+
+        ===============================================================
+
+        Args:
+            schedule: A `str`. The noise schedule of the forward SDE. 'discrete' for discrete-time DPMs,
+                    'linear' or 'cosine' for continuous-time DPMs.
+        Returns:
+            A wrapper object of the forward SDE (VP type).
+        
+        ===============================================================
+
+        Example:
+
+        # For discrete-time DPMs, given betas (the beta array for n = 0, 1, ..., N - 1):
+        >>> ns = NoiseScheduleVP('discrete', betas=betas)
+
+        # For discrete-time DPMs, given alphas_cumprod (the \hat{alpha_n} array for n = 0, 1, ..., N - 1):
+        >>> ns = NoiseScheduleVP('discrete', alphas_cumprod=alphas_cumprod)
+
+        # For continuous-time DPMs (VPSDE), linear schedule:
+        >>> ns = NoiseScheduleVP('linear', continuous_beta_0=0.1, continuous_beta_1=20.)
+
+        """
+
+        if schedule not in ['discrete', 'linear', 'cosine']:
+            raise ValueError("Unsupported noise schedule {}. The schedule needs to be 'discrete' or 'linear' or 'cosine'".format(schedule))
+
+        self.schedule = schedule
+        if schedule == 'discrete':
+            if betas is not None:
+                log_alphas = 0.5 * torch.log(1 - betas).cumsum(dim=0)
+            else:
+                assert alphas_cumprod is not None
+                log_alphas = 0.5 * torch.log(alphas_cumprod)
+            self.total_N = len(log_alphas)
+            self.T = 1.
+            self.t_array = torch.linspace(0., 1., self.total_N + 1)[1:].reshape((1, -1))
+            self.log_alpha_array = log_alphas.reshape((1, -1,))
+        else:
+            self.total_N = 1000
+            self.beta_0 = continuous_beta_0
+            self.beta_1 = continuous_beta_1
+            self.cosine_s = 0.008
+            self.cosine_beta_max = 999.
+            self.cosine_t_max = math.atan(self.cosine_beta_max * (1. + self.cosine_s) / math.pi) * 2. * (1. + self.cosine_s) / math.pi - self.cosine_s
+            self.cosine_log_alpha_0 = math.log(math.cos(self.cosine_s / (1. + self.cosine_s) * math.pi / 2.))
+            self.schedule = schedule
+            if schedule == 'cosine':
+                # For the cosine schedule, T = 1 will have numerical issues. So we manually set the ending time T.
+                # Note that T = 0.9946 may be not the optimal setting. However, we find it works well.
+                self.T = 0.9946
+            else:
+                self.T = 1.
+
+    def marginal_log_mean_coeff(self, t):
+        """
+        Compute log(alpha_t) of a given continuous-time label t in [0, T].
+        """
+        if self.schedule == 'discrete':
+            return interpolate_fn(t.reshape((-1, 1)), self.t_array.to(t.device), self.log_alpha_array.to(t.device)).reshape((-1))
+        elif self.schedule == 'linear':
+            return -0.25 * t ** 2 * (self.beta_1 - self.beta_0) - 0.5 * t * self.beta_0
+        elif self.schedule == 'cosine':
+            log_alpha_fn = lambda s: torch.log(torch.cos((s + self.cosine_s) / (1. + self.cosine_s) * math.pi / 2.))
+            log_alpha_t =  log_alpha_fn(t) - self.cosine_log_alpha_0
+            return log_alpha_t
+
+    def marginal_alpha(self, t):
+        """
+        Compute alpha_t of a given continuous-time label t in [0, T].
+        """
+        return torch.exp(self.marginal_log_mean_coeff(t))
+
+    def marginal_std(self, t):
+        """
+        Compute sigma_t of a given continuous-time label t in [0, T].
+        """
+        return torch.sqrt(1. - torch.exp(2. * self.marginal_log_mean_coeff(t)))
+
+    def marginal_lambda(self, t):
+        """
+        Compute lambda_t = log(alpha_t) - log(sigma_t) of a given continuous-time label t in [0, T].
+        """
+        log_mean_coeff = self.marginal_log_mean_coeff(t)
+        log_std = 0.5 * torch.log(1. - torch.exp(2. * log_mean_coeff))
+        return log_mean_coeff - log_std
+
+    def inverse_lambda(self, lamb):
+        """
+        Compute the continuous-time label t in [0, T] of a given half-logSNR lambda_t.
+        """
+        if self.schedule == 'linear':
+            tmp = 2. * (self.beta_1 - self.beta_0) * torch.logaddexp(-2. * lamb, torch.zeros((1,)).to(lamb))
+            Delta = self.beta_0**2 + tmp
+            return tmp / (torch.sqrt(Delta) + self.beta_0) / (self.beta_1 - self.beta_0)
+        elif self.schedule == 'discrete':
+            log_alpha = -0.5 * torch.logaddexp(torch.zeros((1,)).to(lamb.device), -2. * lamb)
+            t = interpolate_fn(log_alpha.reshape((-1, 1)), torch.flip(self.log_alpha_array.to(lamb.device), [1]), torch.flip(self.t_array.to(lamb.device), [1]))
+            return t.reshape((-1,))
+        else:
+            log_alpha = -0.5 * torch.logaddexp(-2. * lamb, torch.zeros((1,)).to(lamb))
+            t_fn = lambda log_alpha_t: torch.arccos(torch.exp(log_alpha_t + self.cosine_log_alpha_0)) * 2. * (1. + self.cosine_s) / math.pi - self.cosine_s
+            t = t_fn(log_alpha)
+            return t
+
+
+def model_wrapper(
+    model,
+    noise_schedule,
+    model_type="noise",
+    model_kwargs={},
+    guidance_type="uncond",
+    condition=None,
+    unconditional_condition=None,
+    guidance_scale=1.,
+    classifier_fn=None,
+    classifier_kwargs={},
+):
+    """Create a wrapper function for the noise prediction model.
+
+    DPM-Solver needs to solve the continuous-time diffusion ODEs. For DPMs trained on discrete-time labels, we need to
+    firstly wrap the model function to a noise prediction model that accepts the continuous time as the input.
+
+    We support four types of the diffusion model by setting `model_type`:
+
+        1. "noise": noise prediction model. (Trained by predicting noise).
+
+        2. "x_start": data prediction model. (Trained by predicting the data x_0 at time 0).
+
+        3. "v": velocity prediction model. (Trained by predicting the velocity).
+            The "v" prediction is derivation detailed in Appendix D of [1], and is used in Imagen-Video [2].
+
+            [1] Salimans, Tim, and Jonathan Ho. "Progressive distillation for fast sampling of diffusion models."
+                arXiv preprint arXiv:2202.00512 (2022).
+            [2] Ho, Jonathan, et al. "Imagen Video: High Definition Video Generation with Diffusion Models."
+                arXiv preprint arXiv:2210.02303 (2022).
+    
+        4. "score": marginal score function. (Trained by denoising score matching).
+            Note that the score function and the noise prediction model follows a simple relationship:
+            ```
+                noise(x_t, t) = -sigma_t * score(x_t, t)
+            ```
+
+    We support three types of guided sampling by DPMs by setting `guidance_type`:
+        1. "uncond": unconditional sampling by DPMs.
+            The input `model` has the following format:
+            ``
+                model(x, t_input, **model_kwargs) -> noise | x_start | v | score
+            ``
+
+        2. "classifier": classifier guidance sampling [3] by DPMs and another classifier.
+            The input `model` has the following format:
+            ``
+                model(x, t_input, **model_kwargs) -> noise | x_start | v | score
+            `` 
+
+            The input `classifier_fn` has the following format:
+            ``
+                classifier_fn(x, t_input, cond, **classifier_kwargs) -> logits(x, t_input, cond)
+            ``
+
+            [3] P. Dhariwal and A. Q. Nichol, "Diffusion models beat GANs on image synthesis,"
+                in Advances in Neural Information Processing Systems, vol. 34, 2021, pp. 8780-8794.
+
+        3. "classifier-free": classifier-free guidance sampling by conditional DPMs.
+            The input `model` has the following format:
+            ``
+                model(x, t_input, cond, **model_kwargs) -> noise | x_start | v | score
+            `` 
+            And if cond == `unconditional_condition`, the model output is the unconditional DPM output.
+
+            [4] Ho, Jonathan, and Tim Salimans. "Classifier-free diffusion guidance."
+                arXiv preprint arXiv:2207.12598 (2022).
+        
+
+    The `t_input` is the time label of the model, which may be discrete-time labels (i.e. 0 to 999)
+    or continuous-time labels (i.e. epsilon to T).
+
+    We wrap the model function to accept only `x` and `t_continuous` as inputs, and outputs the predicted noise:
+    ``
+        def model_fn(x, t_continuous) -> noise:
+            t_input = get_model_input_time(t_continuous)
+            return noise_pred(model, x, t_input, **model_kwargs)         
+    ``
+    where `t_continuous` is the continuous time labels (i.e. epsilon to T). And we use `model_fn` for DPM-Solver.
+
+    ===============================================================
+
+    Args:
+        model: A diffusion model with the corresponding format described above.
+        noise_schedule: A noise schedule object, such as NoiseScheduleVP.
+        model_type: A `str`. The parameterization type of the diffusion model.
+                    "noise" or "x_start" or "v" or "score".
+        model_kwargs: A `dict`. A dict for the other inputs of the model function.
+        guidance_type: A `str`. The type of the guidance for sampling.
+                    "uncond" or "classifier" or "classifier-free".
+        condition: A pytorch tensor. The condition for the guided sampling.
+                    Only used for "classifier" or "classifier-free" guidance type.
+        unconditional_condition: A pytorch tensor. The condition for the unconditional sampling.
+                    Only used for "classifier-free" guidance type.
+        guidance_scale: A `float`. The scale for the guided sampling.
+        classifier_fn: A classifier function. Only used for the classifier guidance.
+        classifier_kwargs: A `dict`. A dict for the other inputs of the classifier function.
+    Returns:
+        A noise prediction model that accepts the noised data and the continuous time as the inputs.
+    """
+
+    def get_model_input_time(t_continuous):
+        """
+        Convert the continuous-time `t_continuous` (in [epsilon, T]) to the model input time.
+        For discrete-time DPMs, we convert `t_continuous` in [1 / N, 1] to `t_input` in [0, 1000 * (N - 1) / N].
+        For continuous-time DPMs, we just use `t_continuous`.
+        """
+        if noise_schedule.schedule == 'discrete':
+            return (t_continuous - 1. / noise_schedule.total_N) * 1000.
+        else:
+            return t_continuous
+
+    def noise_pred_fn(x, t_continuous, cond=None):
+        if t_continuous.reshape((-1,)).shape[0] == 1:
+            t_continuous = t_continuous.expand((x.shape[0]))
+        t_input = get_model_input_time(t_continuous)
+        output = model(x, t_input, **model_kwargs)
+        if model_type == "noise":
+            return output
+        elif model_type == "x_start":
+            alpha_t, sigma_t = noise_schedule.marginal_alpha(t_continuous), noise_schedule.marginal_std(t_continuous)
+            dims = x.dim()
+            return (x - expand_dims(alpha_t, dims) * output) / expand_dims(sigma_t, dims)
+        elif model_type == "v":
+            alpha_t, sigma_t = noise_schedule.marginal_alpha(t_continuous), noise_schedule.marginal_std(t_continuous)
+            dims = x.dim()
+            return expand_dims(alpha_t, dims) * output + expand_dims(sigma_t, dims) * x
+        elif model_type == "score":
+            sigma_t = noise_schedule.marginal_std(t_continuous)
+            dims = x.dim()
+            return -expand_dims(sigma_t, dims) * output
+
+    def cond_grad_fn(x, t_input):
+        """
+        Compute the gradient of the classifier, i.e. nabla_{x} log p_t(cond | x_t).
+        """
+        with torch.enable_grad():
+            x_in = x.detach().requires_grad_(True)
+            log_prob = classifier_fn(x_in, t_input, condition, **classifier_kwargs)
+            return torch.autograd.grad(log_prob.sum(), x_in)[0]
+
+    def model_fn(x, t_continuous):
+        """
+        The noise predicition model function that is used for DPM-Solver.
+        """
+        if t_continuous.reshape((-1,)).shape[0] == 1:
+            t_continuous = t_continuous.expand((x.shape[0]))
+        if guidance_type == "uncond":
+            return noise_pred_fn(x, t_continuous)
+        elif guidance_type == "classifier":
+            assert classifier_fn is not None
+            t_input = get_model_input_time(t_continuous)
+            cond_grad = cond_grad_fn(x, t_input)
+            sigma_t = noise_schedule.marginal_std(t_continuous)
+            noise = noise_pred_fn(x, t_continuous)
+            return noise - guidance_scale * expand_dims(sigma_t, dims=cond_grad.dim()) * cond_grad
+        elif guidance_type == "classifier-free":
+            if guidance_scale == 1. or unconditional_condition is None:
+                return noise_pred_fn(x, t_continuous, cond=condition)
+            else:
+                x_in = torch.cat([x] * 2)
+                t_in = torch.cat([t_continuous] * 2)
+                c_in = torch.cat([unconditional_condition, condition])
+                noise_uncond, noise = noise_pred_fn(x_in, t_in, cond=c_in).chunk(2)
+                return noise_uncond + guidance_scale * (noise - noise_uncond)
+
+    assert model_type in ["noise", "x_start", "v"]
+    assert guidance_type in ["uncond", "classifier", "classifier-free"]
+    return model_fn
+
+
+class UniPC:
+    def __init__(
+        self,
+        model_fn,
+        noise_schedule,
+        predict_x0=True,
+        thresholding=False,
+        max_val=1.,
+        variant='bh1',
+        noise_mask=None,
+        masked_image=None,
+        noise=None,
+    ):
+        """Construct a UniPC. 
+
+        We support both data_prediction and noise_prediction.
+        """
+        self.model = model_fn
+        self.noise_schedule = noise_schedule
+        self.variant = variant
+        self.predict_x0 = predict_x0
+        self.thresholding = thresholding
+        self.max_val = max_val
+        self.noise_mask = noise_mask
+        self.masked_image = masked_image
+        self.noise = noise
+
+    def dynamic_thresholding_fn(self, x0, t=None):
+        """
+        The dynamic thresholding method. 
+        """
+        dims = x0.dim()
+        p = self.dynamic_thresholding_ratio
+        s = torch.quantile(torch.abs(x0).reshape((x0.shape[0], -1)), p, dim=1)
+        s = expand_dims(torch.maximum(s, self.thresholding_max_val * torch.ones_like(s).to(s.device)), dims)
+        x0 = torch.clamp(x0, -s, s) / s
+        return x0
+
+    def noise_prediction_fn(self, x, t):
+        """
+        Return the noise prediction model.
+        """
+        if self.noise_mask is not None:
+            return self.model(x, t) * self.noise_mask
+        else:
+            return self.model(x, t)
+
+    def data_prediction_fn(self, x, t):
+        """
+        Return the data prediction model (with thresholding).
+        """
+        noise = self.noise_prediction_fn(x, t)
+        dims = x.dim()
+        alpha_t, sigma_t = self.noise_schedule.marginal_alpha(t), self.noise_schedule.marginal_std(t)
+        x0 = (x - expand_dims(sigma_t, dims) * noise) / expand_dims(alpha_t, dims)
+        if self.thresholding:
+            p = 0.995   # A hyperparameter in the paper of "Imagen" [1].
+            s = torch.quantile(torch.abs(x0).reshape((x0.shape[0], -1)), p, dim=1)
+            s = expand_dims(torch.maximum(s, self.max_val * torch.ones_like(s).to(s.device)), dims)
+            x0 = torch.clamp(x0, -s, s) / s
+        if self.noise_mask is not None:
+            x0 = x0 * self.noise_mask + (1. - self.noise_mask) * self.masked_image
+        return x0
+
+    def model_fn(self, x, t):
+        """
+        Convert the model to the noise prediction model or the data prediction model. 
+        """
+        if self.predict_x0:
+            return self.data_prediction_fn(x, t)
+        else:
+            return self.noise_prediction_fn(x, t)
+
+    def get_time_steps(self, skip_type, t_T, t_0, N, device):
+        """Compute the intermediate time steps for sampling.
+        """
+        if skip_type == 'logSNR':
+            lambda_T = self.noise_schedule.marginal_lambda(torch.tensor(t_T).to(device))
+            lambda_0 = self.noise_schedule.marginal_lambda(torch.tensor(t_0).to(device))
+            logSNR_steps = torch.linspace(lambda_T.cpu().item(), lambda_0.cpu().item(), N + 1).to(device)
+            return self.noise_schedule.inverse_lambda(logSNR_steps)
+        elif skip_type == 'time_uniform':
+            return torch.linspace(t_T, t_0, N + 1).to(device)
+        elif skip_type == 'time_quadratic':
+            t_order = 2
+            t = torch.linspace(t_T**(1. / t_order), t_0**(1. / t_order), N + 1).pow(t_order).to(device)
+            return t
+        else:
+            raise ValueError("Unsupported skip_type {}, need to be 'logSNR' or 'time_uniform' or 'time_quadratic'".format(skip_type))
+
+    def get_orders_and_timesteps_for_singlestep_solver(self, steps, order, skip_type, t_T, t_0, device):
+        """
+        Get the order of each step for sampling by the singlestep DPM-Solver.
+        """
+        if order == 3:
+            K = steps // 3 + 1
+            if steps % 3 == 0:
+                orders = [3,] * (K - 2) + [2, 1]
+            elif steps % 3 == 1:
+                orders = [3,] * (K - 1) + [1]
+            else:
+                orders = [3,] * (K - 1) + [2]
+        elif order == 2:
+            if steps % 2 == 0:
+                K = steps // 2
+                orders = [2,] * K
+            else:
+                K = steps // 2 + 1
+                orders = [2,] * (K - 1) + [1]
+        elif order == 1:
+            K = steps
+            orders = [1,] * steps
+        else:
+            raise ValueError("'order' must be '1' or '2' or '3'.")
+        if skip_type == 'logSNR':
+            # To reproduce the results in DPM-Solver paper
+            timesteps_outer = self.get_time_steps(skip_type, t_T, t_0, K, device)
+        else:
+            timesteps_outer = self.get_time_steps(skip_type, t_T, t_0, steps, device)[torch.cumsum(torch.tensor([0,] + orders), 0).to(device)]
+        return timesteps_outer, orders
+
+    def denoise_to_zero_fn(self, x, s):
+        """
+        Denoise at the final step, which is equivalent to solve the ODE from lambda_s to infty by first-order discretization. 
+        """
+        return self.data_prediction_fn(x, s)
+
+    def multistep_uni_pc_update(self, x, model_prev_list, t_prev_list, t, order, **kwargs):
+        if len(t.shape) == 0:
+            t = t.view(-1)
+        if 'bh' in self.variant:
+            return self.multistep_uni_pc_bh_update(x, model_prev_list, t_prev_list, t, order, **kwargs)
+        else:
+            assert self.variant == 'vary_coeff'
+            return self.multistep_uni_pc_vary_update(x, model_prev_list, t_prev_list, t, order, **kwargs)
+
+    def multistep_uni_pc_vary_update(self, x, model_prev_list, t_prev_list, t, order, use_corrector=True):
+        print(f'using unified predictor-corrector with order {order} (solver type: vary coeff)')
+        ns = self.noise_schedule
+        assert order <= len(model_prev_list)
+
+        # first compute rks
+        t_prev_0 = t_prev_list[-1]
+        lambda_prev_0 = ns.marginal_lambda(t_prev_0)
+        lambda_t = ns.marginal_lambda(t)
+        model_prev_0 = model_prev_list[-1]
+        sigma_prev_0, sigma_t = ns.marginal_std(t_prev_0), ns.marginal_std(t)
+        log_alpha_t = ns.marginal_log_mean_coeff(t)
+        alpha_t = torch.exp(log_alpha_t)
+
+        h = lambda_t - lambda_prev_0
+
+        rks = []
+        D1s = []
+        for i in range(1, order):
+            t_prev_i = t_prev_list[-(i + 1)]
+            model_prev_i = model_prev_list[-(i + 1)]
+            lambda_prev_i = ns.marginal_lambda(t_prev_i)
+            rk = (lambda_prev_i - lambda_prev_0) / h
+            rks.append(rk)
+            D1s.append((model_prev_i - model_prev_0) / rk)
+
+        rks.append(1.)
+        rks = torch.tensor(rks, device=x.device)
+
+        K = len(rks)
+        # build C matrix
+        C = []
+
+        col = torch.ones_like(rks)
+        for k in range(1, K + 1):
+            C.append(col)
+            col = col * rks / (k + 1) 
+        C = torch.stack(C, dim=1)
+
+        if len(D1s) > 0:
+            D1s = torch.stack(D1s, dim=1) # (B, K)
+            C_inv_p = torch.linalg.inv(C[:-1, :-1])
+            A_p = C_inv_p
+
+        if use_corrector:
+            print('using corrector')
+            C_inv = torch.linalg.inv(C)
+            A_c = C_inv
+
+        hh = -h if self.predict_x0 else h
+        h_phi_1 = torch.expm1(hh)
+        h_phi_ks = []
+        factorial_k = 1
+        h_phi_k = h_phi_1
+        for k in range(1, K + 2):
+            h_phi_ks.append(h_phi_k)
+            h_phi_k = h_phi_k / hh - 1 / factorial_k
+            factorial_k *= (k + 1)
+
+        model_t = None
+        if self.predict_x0:
+            x_t_ = (
+                sigma_t / sigma_prev_0 * x
+                - alpha_t * h_phi_1 * model_prev_0
+            )
+            # now predictor
+            x_t = x_t_
+            if len(D1s) > 0:
+                # compute the residuals for predictor
+                for k in range(K - 1):
+                    x_t = x_t - alpha_t * h_phi_ks[k + 1] * torch.einsum('bkchw,k->bchw', D1s, A_p[k])
+            # now corrector
+            if use_corrector:
+                model_t = self.model_fn(x_t, t)
+                D1_t = (model_t - model_prev_0)
+                x_t = x_t_
+                k = 0
+                for k in range(K - 1):
+                    x_t = x_t - alpha_t * h_phi_ks[k + 1] * torch.einsum('bkchw,k->bchw', D1s, A_c[k][:-1])
+                x_t = x_t - alpha_t * h_phi_ks[K] * (D1_t * A_c[k][-1])
+        else:
+            log_alpha_prev_0, log_alpha_t = ns.marginal_log_mean_coeff(t_prev_0), ns.marginal_log_mean_coeff(t)
+            x_t_ = (
+                (torch.exp(log_alpha_t - log_alpha_prev_0)) * x
+                - (sigma_t * h_phi_1) * model_prev_0
+            )
+            # now predictor
+            x_t = x_t_
+            if len(D1s) > 0:
+                # compute the residuals for predictor
+                for k in range(K - 1):
+                    x_t = x_t - sigma_t * h_phi_ks[k + 1] * torch.einsum('bkchw,k->bchw', D1s, A_p[k])
+            # now corrector
+            if use_corrector:
+                model_t = self.model_fn(x_t, t)
+                D1_t = (model_t - model_prev_0)
+                x_t = x_t_
+                k = 0
+                for k in range(K - 1):
+                    x_t = x_t - sigma_t * h_phi_ks[k + 1] * torch.einsum('bkchw,k->bchw', D1s, A_c[k][:-1])
+                x_t = x_t - sigma_t * h_phi_ks[K] * (D1_t * A_c[k][-1])
+        return x_t, model_t
+
+    def multistep_uni_pc_bh_update(self, x, model_prev_list, t_prev_list, t, order, x_t=None, use_corrector=True):
+        # print(f'using unified predictor-corrector with order {order} (solver type: B(h))')
+        ns = self.noise_schedule
+        assert order <= len(model_prev_list)
+        dims = x.dim()
+
+        # first compute rks
+        t_prev_0 = t_prev_list[-1]
+        lambda_prev_0 = ns.marginal_lambda(t_prev_0)
+        lambda_t = ns.marginal_lambda(t)
+        model_prev_0 = model_prev_list[-1]
+        sigma_prev_0, sigma_t = ns.marginal_std(t_prev_0), ns.marginal_std(t)
+        log_alpha_prev_0, log_alpha_t = ns.marginal_log_mean_coeff(t_prev_0), ns.marginal_log_mean_coeff(t)
+        alpha_t = torch.exp(log_alpha_t)
+
+        h = lambda_t - lambda_prev_0
+
+        rks = []
+        D1s = []
+        for i in range(1, order):
+            t_prev_i = t_prev_list[-(i + 1)]
+            model_prev_i = model_prev_list[-(i + 1)]
+            lambda_prev_i = ns.marginal_lambda(t_prev_i)
+            rk = ((lambda_prev_i - lambda_prev_0) / h)[0]
+            rks.append(rk)
+            D1s.append((model_prev_i - model_prev_0) / rk)
+
+        rks.append(1.)
+        rks = torch.tensor(rks, device=x.device)
+
+        R = []
+        b = []
+
+        hh = -h[0] if self.predict_x0 else h[0]
+        h_phi_1 = torch.expm1(hh) # h\phi_1(h) = e^h - 1
+        h_phi_k = h_phi_1 / hh - 1
+
+        factorial_i = 1
+
+        if self.variant == 'bh1':
+            B_h = hh
+        elif self.variant == 'bh2':
+            B_h = torch.expm1(hh)
+        else:
+            raise NotImplementedError()
+            
+        for i in range(1, order + 1):
+            R.append(torch.pow(rks, i - 1))
+            b.append(h_phi_k * factorial_i / B_h)
+            factorial_i *= (i + 1)
+            h_phi_k = h_phi_k / hh - 1 / factorial_i 
+
+        R = torch.stack(R)
+        b = torch.tensor(b, device=x.device)
+
+        # now predictor
+        use_predictor = len(D1s) > 0 and x_t is None
+        if len(D1s) > 0:
+            D1s = torch.stack(D1s, dim=1) # (B, K)
+            if x_t is None:
+                # for order 2, we use a simplified version
+                if order == 2:
+                    rhos_p = torch.tensor([0.5], device=b.device)
+                else:
+                    rhos_p = torch.linalg.solve(R[:-1, :-1], b[:-1])
+        else:
+            D1s = None
+
+        if use_corrector:
+            # print('using corrector')
+            # for order 1, we use a simplified version
+            if order == 1:
+                rhos_c = torch.tensor([0.5], device=b.device)
+            else:
+                rhos_c = torch.linalg.solve(R, b)
+
+        model_t = None
+        if self.predict_x0:
+            x_t_ = (
+                expand_dims(sigma_t / sigma_prev_0, dims) * x
+                - expand_dims(alpha_t * h_phi_1, dims)* model_prev_0
+            )
+
+            if x_t is None:
+                if use_predictor:
+                    pred_res = torch.einsum('k,bkchw->bchw', rhos_p, D1s)
+                else:
+                    pred_res = 0
+                x_t = x_t_ - expand_dims(alpha_t * B_h, dims) * pred_res
+
+            if use_corrector:
+                model_t = self.model_fn(x_t, t)
+                if D1s is not None:
+                    corr_res = torch.einsum('k,bkchw->bchw', rhos_c[:-1], D1s)
+                else:
+                    corr_res = 0
+                D1_t = (model_t - model_prev_0)
+                x_t = x_t_ - expand_dims(alpha_t * B_h, dims) * (corr_res + rhos_c[-1] * D1_t)
+        else:
+            x_t_ = (
+                expand_dims(torch.exp(log_alpha_t - log_alpha_prev_0), dims) * x
+                - expand_dims(sigma_t * h_phi_1, dims) * model_prev_0
+            )
+            if x_t is None:
+                if use_predictor:
+                    pred_res = torch.einsum('k,bkchw->bchw', rhos_p, D1s)
+                else:
+                    pred_res = 0
+                x_t = x_t_ - expand_dims(sigma_t * B_h, dims) * pred_res
+
+            if use_corrector:
+                model_t = self.model_fn(x_t, t)
+                if D1s is not None:
+                    corr_res = torch.einsum('k,bkchw->bchw', rhos_c[:-1], D1s)
+                else:
+                    corr_res = 0
+                D1_t = (model_t - model_prev_0)
+                x_t = x_t_ - expand_dims(sigma_t * B_h, dims) * (corr_res + rhos_c[-1] * D1_t)
+        return x_t, model_t
+
+
+    def sample(self, x, timesteps, t_start=None, t_end=None, order=3, skip_type='time_uniform',
+        method='singlestep', lower_order_final=True, denoise_to_zero=False, solver_type='dpm_solver',
+        atol=0.0078, rtol=0.05, corrector=False, callback=None, disable_pbar=False
+    ):
+        # t_0 = 1. / self.noise_schedule.total_N if t_end is None else t_end
+        # t_T = self.noise_schedule.T if t_start is None else t_start
+        device = x.device
+        steps = len(timesteps) - 1
+        if method == 'multistep':
+            assert steps >= order
+            # timesteps = self.get_time_steps(skip_type=skip_type, t_T=t_T, t_0=t_0, N=steps, device=device)
+            assert timesteps.shape[0] - 1 == steps
+            # with torch.no_grad():
+            for step_index in trange(steps, disable=disable_pbar):
+                if self.noise_mask is not None:
+                    x = x * self.noise_mask + (1. - self.noise_mask) * (self.masked_image * self.noise_schedule.marginal_alpha(timesteps[step_index]) + self.noise * self.noise_schedule.marginal_std(timesteps[step_index]))
+                if step_index == 0:
+                    vec_t = timesteps[0].expand((x.shape[0]))
+                    model_prev_list = [self.model_fn(x, vec_t)]
+                    t_prev_list = [vec_t]
+                elif step_index < order:
+                    init_order = step_index
+                # Init the first `order` values by lower order multistep DPM-Solver.
+                # for init_order in range(1, order):
+                    vec_t = timesteps[init_order].expand(x.shape[0])
+                    x, model_x = self.multistep_uni_pc_update(x, model_prev_list, t_prev_list, vec_t, init_order, use_corrector=True)
+                    if model_x is None:
+                        model_x = self.model_fn(x, vec_t)
+                    model_prev_list.append(model_x)
+                    t_prev_list.append(vec_t)
+                else:
+                    extra_final_step = 0
+                    if step_index == (steps - 1):
+                        extra_final_step = 1
+                    for step in range(step_index, step_index + 1 + extra_final_step):
+                        vec_t = timesteps[step].expand(x.shape[0])
+                        if lower_order_final:
+                            step_order = min(order, steps + 1 - step)
+                        else:
+                            step_order = order
+                        # print('this step order:', step_order)
+                        if step == steps:
+                            # print('do not run corrector at the last step')
+                            use_corrector = False
+                        else:
+                            use_corrector = True
+                        x, model_x =  self.multistep_uni_pc_update(x, model_prev_list, t_prev_list, vec_t, step_order, use_corrector=use_corrector)
+                        for i in range(order - 1):
+                            t_prev_list[i] = t_prev_list[i + 1]
+                            model_prev_list[i] = model_prev_list[i + 1]
+                        t_prev_list[-1] = vec_t
+                        # We do not need to evaluate the final model value.
+                        if step < steps:
+                            if model_x is None:
+                                model_x = self.model_fn(x, vec_t)
+                            model_prev_list[-1] = model_x
+                if callback is not None:
+                    callback(step_index, model_prev_list[-1], x, steps)
+        else:
+            raise NotImplementedError()
+        # if denoise_to_zero:
+        #     x = self.denoise_to_zero_fn(x, torch.ones((x.shape[0],)).to(device) * t_0)
+        return x
+
+
+#############################################################
+# other utility functions
+#############################################################
+
+def interpolate_fn(x, xp, yp):
+    """
+    A piecewise linear function y = f(x), using xp and yp as keypoints.
+    We implement f(x) in a differentiable way (i.e. applicable for autograd).
+    The function f(x) is well-defined for all x-axis. (For x beyond the bounds of xp, we use the outmost points of xp to define the linear function.)
+
+    Args:
+        x: PyTorch tensor with shape [N, C], where N is the batch size, C is the number of channels (we use C = 1 for DPM-Solver).
+        xp: PyTorch tensor with shape [C, K], where K is the number of keypoints.
+        yp: PyTorch tensor with shape [C, K].
+    Returns:
+        The function values f(x), with shape [N, C].
+    """
+    N, K = x.shape[0], xp.shape[1]
+    all_x = torch.cat([x.unsqueeze(2), xp.unsqueeze(0).repeat((N, 1, 1))], dim=2)
+    sorted_all_x, x_indices = torch.sort(all_x, dim=2)
+    x_idx = torch.argmin(x_indices, dim=2)
+    cand_start_idx = x_idx - 1
+    start_idx = torch.where(
+        torch.eq(x_idx, 0),
+        torch.tensor(1, device=x.device),
+        torch.where(
+            torch.eq(x_idx, K), torch.tensor(K - 2, device=x.device), cand_start_idx,
+        ),
+    )
+    end_idx = torch.where(torch.eq(start_idx, cand_start_idx), start_idx + 2, start_idx + 1)
+    start_x = torch.gather(sorted_all_x, dim=2, index=start_idx.unsqueeze(2)).squeeze(2)
+    end_x = torch.gather(sorted_all_x, dim=2, index=end_idx.unsqueeze(2)).squeeze(2)
+    start_idx2 = torch.where(
+        torch.eq(x_idx, 0),
+        torch.tensor(0, device=x.device),
+        torch.where(
+            torch.eq(x_idx, K), torch.tensor(K - 2, device=x.device), cand_start_idx,
+        ),
+    )
+    y_positions_expanded = yp.unsqueeze(0).expand(N, -1, -1)
+    start_y = torch.gather(y_positions_expanded, dim=2, index=start_idx2.unsqueeze(2)).squeeze(2)
+    end_y = torch.gather(y_positions_expanded, dim=2, index=(start_idx2 + 1).unsqueeze(2)).squeeze(2)
+    cand = start_y + (x - start_x) * (end_y - start_y) / (end_x - start_x)
+    return cand
+
+
+def expand_dims(v, dims):
+    """
+    Expand the tensor `v` to the dim `dims`.
+
+    Args:
+        `v`: a PyTorch tensor with shape [N].
+        `dim`: a `int`.
+    Returns:
+        a PyTorch tensor with shape [N, 1, 1, ..., 1] and the total dimension is `dims`.
+    """
+    return v[(...,) + (None,)*(dims - 1)]
+
+
+class SigmaConvert:
+    schedule = ""
+    def marginal_log_mean_coeff(self, sigma):
+        return 0.5 * torch.log(1 / ((sigma * sigma) + 1))
+
+    def marginal_alpha(self, t):
+        return torch.exp(self.marginal_log_mean_coeff(t))
+
+    def marginal_std(self, t):
+        return torch.sqrt(1. - torch.exp(2. * self.marginal_log_mean_coeff(t)))
+
+    def marginal_lambda(self, t):
+        """
+        Compute lambda_t = log(alpha_t) - log(sigma_t) of a given continuous-time label t in [0, T].
+        """
+        log_mean_coeff = self.marginal_log_mean_coeff(t)
+        log_std = 0.5 * torch.log(1. - torch.exp(2. * log_mean_coeff))
+        return log_mean_coeff - log_std
+
+def predict_eps_sigma(model, input, sigma_in, **kwargs):
+    sigma = sigma_in.view(sigma_in.shape[:1] + (1,) * (input.ndim - 1))
+    input = input * ((sigma ** 2 + 1.0) ** 0.5)
+    return  (input - model(input, sigma_in, **kwargs)) / sigma
+
+
+def sample_unipc(model, noise, image, sigmas, sampling_function, max_denoise, extra_args=None, callback=None, disable=False, noise_mask=None, variant='bh1'):
+        timesteps = sigmas.clone()
+        if sigmas[-1] == 0:
+            timesteps = sigmas[:]
+            timesteps[-1] = 0.001
+        else:
+            timesteps = sigmas.clone()
+        ns = SigmaConvert()
+
+        if image is not None:
+            img = image * ns.marginal_alpha(timesteps[0])
+            if max_denoise:
+                noise_mult = 1.0
+            else:
+                noise_mult = ns.marginal_std(timesteps[0])
+            img += noise * noise_mult
+        else:
+            img = noise
+
+        model_type = "noise"
+
+        model_fn = model_wrapper(
+            lambda input, sigma, **kwargs: predict_eps_sigma(model, input, sigma, **kwargs),
+            ns,
+            model_type=model_type,
+            guidance_type="uncond",
+            model_kwargs=extra_args,
+        )
+
+        order = min(3, len(timesteps) - 2)
+        uni_pc = UniPC(model_fn, ns, predict_x0=True, thresholding=False, noise_mask=noise_mask, masked_image=image, noise=noise, variant=variant)
+        x = uni_pc.sample(img, timesteps=timesteps, skip_type="time_uniform", method="multistep", order=order, lower_order_final=True, callback=callback, disable_pbar=disable)
+        x /= ns.marginal_alpha(timesteps[-1])
+        return x

backend/headless/fcbh/gligen.py

@@ -0,0 +1,341 @@
+import torch
+from torch import nn, einsum
+from .ldm.modules.attention import CrossAttention
+from inspect import isfunction
+
+
+def exists(val):
+    return val is not None
+
+
+def uniq(arr):
+    return{el: True for el in arr}.keys()
+
+
+def default(val, d):
+    if exists(val):
+        return val
+    return d() if isfunction(d) else d
+
+
+# feedforward
+class GEGLU(nn.Module):
+    def __init__(self, dim_in, dim_out):
+        super().__init__()
+        self.proj = nn.Linear(dim_in, dim_out * 2)
+
+    def forward(self, x):
+        x, gate = self.proj(x).chunk(2, dim=-1)
+        return x * torch.nn.functional.gelu(gate)
+
+
+class FeedForward(nn.Module):
+    def __init__(self, dim, dim_out=None, mult=4, glu=False, dropout=0.):
+        super().__init__()
+        inner_dim = int(dim * mult)
+        dim_out = default(dim_out, dim)
+        project_in = nn.Sequential(
+            nn.Linear(dim, inner_dim),
+            nn.GELU()
+        ) if not glu else GEGLU(dim, inner_dim)
+
+        self.net = nn.Sequential(
+            project_in,
+            nn.Dropout(dropout),
+            nn.Linear(inner_dim, dim_out)
+        )
+
+    def forward(self, x):
+        return self.net(x)
+
+
+class GatedCrossAttentionDense(nn.Module):
+    def __init__(self, query_dim, context_dim, n_heads, d_head):
+        super().__init__()
+
+        self.attn = CrossAttention(
+            query_dim=query_dim,
+            context_dim=context_dim,
+            heads=n_heads,
+            dim_head=d_head)
+        self.ff = FeedForward(query_dim, glu=True)
+
+        self.norm1 = nn.LayerNorm(query_dim)
+        self.norm2 = nn.LayerNorm(query_dim)
+
+        self.register_parameter('alpha_attn', nn.Parameter(torch.tensor(0.)))
+        self.register_parameter('alpha_dense', nn.Parameter(torch.tensor(0.)))
+
+        # this can be useful: we can externally change magnitude of tanh(alpha)
+        # for example, when it is set to 0, then the entire model is same as
+        # original one
+        self.scale = 1
+
+    def forward(self, x, objs):
+
+        x = x + self.scale * \
+            torch.tanh(self.alpha_attn) * self.attn(self.norm1(x), objs, objs)
+        x = x + self.scale * \
+            torch.tanh(self.alpha_dense) * self.ff(self.norm2(x))
+
+        return x
+
+
+class GatedSelfAttentionDense(nn.Module):
+    def __init__(self, query_dim, context_dim, n_heads, d_head):
+        super().__init__()
+
+        # we need a linear projection since we need cat visual feature and obj
+        # feature
+        self.linear = nn.Linear(context_dim, query_dim)
+
+        self.attn = CrossAttention(
+            query_dim=query_dim,
+            context_dim=query_dim,
+            heads=n_heads,
+            dim_head=d_head)
+        self.ff = FeedForward(query_dim, glu=True)
+
+        self.norm1 = nn.LayerNorm(query_dim)
+        self.norm2 = nn.LayerNorm(query_dim)
+
+        self.register_parameter('alpha_attn', nn.Parameter(torch.tensor(0.)))
+        self.register_parameter('alpha_dense', nn.Parameter(torch.tensor(0.)))
+
+        # this can be useful: we can externally change magnitude of tanh(alpha)
+        # for example, when it is set to 0, then the entire model is same as
+        # original one
+        self.scale = 1
+
+    def forward(self, x, objs):
+
+        N_visual = x.shape[1]
+        objs = self.linear(objs)
+
+        x = x + self.scale * torch.tanh(self.alpha_attn) * self.attn(
+            self.norm1(torch.cat([x, objs], dim=1)))[:, 0:N_visual, :]
+        x = x + self.scale * \
+            torch.tanh(self.alpha_dense) * self.ff(self.norm2(x))
+
+        return x
+
+
+class GatedSelfAttentionDense2(nn.Module):
+    def __init__(self, query_dim, context_dim, n_heads, d_head):
+        super().__init__()
+
+        # we need a linear projection since we need cat visual feature and obj
+        # feature
+        self.linear = nn.Linear(context_dim, query_dim)
+
+        self.attn = CrossAttention(
+            query_dim=query_dim, context_dim=query_dim, dim_head=d_head)
+        self.ff = FeedForward(query_dim, glu=True)
+
+        self.norm1 = nn.LayerNorm(query_dim)
+        self.norm2 = nn.LayerNorm(query_dim)
+
+        self.register_parameter('alpha_attn', nn.Parameter(torch.tensor(0.)))
+        self.register_parameter('alpha_dense', nn.Parameter(torch.tensor(0.)))
+
+        # this can be useful: we can externally change magnitude of tanh(alpha)
+        # for example, when it is set to 0, then the entire model is same as
+        # original one
+        self.scale = 1
+
+    def forward(self, x, objs):
+
+        B, N_visual, _ = x.shape
+        B, N_ground, _ = objs.shape
+
+        objs = self.linear(objs)
+
+        # sanity check
+        size_v = math.sqrt(N_visual)
+        size_g = math.sqrt(N_ground)
+        assert int(size_v) == size_v, "Visual tokens must be square rootable"
+        assert int(size_g) == size_g, "Grounding tokens must be square rootable"
+        size_v = int(size_v)
+        size_g = int(size_g)
+
+        # select grounding token and resize it to visual token size as residual
+        out = self.attn(self.norm1(torch.cat([x, objs], dim=1)))[
+            :, N_visual:, :]
+        out = out.permute(0, 2, 1).reshape(B, -1, size_g, size_g)
+        out = torch.nn.functional.interpolate(
+            out, (size_v, size_v), mode='bicubic')
+        residual = out.reshape(B, -1, N_visual).permute(0, 2, 1)
+
+        # add residual to visual feature
+        x = x + self.scale * torch.tanh(self.alpha_attn) * residual
+        x = x + self.scale * \
+            torch.tanh(self.alpha_dense) * self.ff(self.norm2(x))
+
+        return x
+
+
+class FourierEmbedder():
+    def __init__(self, num_freqs=64, temperature=100):
+
+        self.num_freqs = num_freqs
+        self.temperature = temperature
+        self.freq_bands = temperature ** (torch.arange(num_freqs) / num_freqs)
+
+    @torch.no_grad()
+    def __call__(self, x, cat_dim=-1):
+        "x: arbitrary shape of tensor. dim: cat dim"
+        out = []
+        for freq in self.freq_bands:
+            out.append(torch.sin(freq * x))
+            out.append(torch.cos(freq * x))
+        return torch.cat(out, cat_dim)
+
+
+class PositionNet(nn.Module):
+    def __init__(self, in_dim, out_dim, fourier_freqs=8):
+        super().__init__()
+        self.in_dim = in_dim
+        self.out_dim = out_dim
+
+        self.fourier_embedder = FourierEmbedder(num_freqs=fourier_freqs)
+        self.position_dim = fourier_freqs * 2 * 4  # 2 is sin&cos, 4 is xyxy
+
+        self.linears = nn.Sequential(
+            nn.Linear(self.in_dim + self.position_dim, 512),
+            nn.SiLU(),
+            nn.Linear(512, 512),
+            nn.SiLU(),
+            nn.Linear(512, out_dim),
+        )
+
+        self.null_positive_feature = torch.nn.Parameter(
+            torch.zeros([self.in_dim]))
+        self.null_position_feature = torch.nn.Parameter(
+            torch.zeros([self.position_dim]))
+
+    def forward(self, boxes, masks, positive_embeddings):
+        B, N, _ = boxes.shape
+        dtype = self.linears[0].weight.dtype
+        masks = masks.unsqueeze(-1).to(dtype)
+        positive_embeddings = positive_embeddings.to(dtype)
+
+        # embedding position (it may includes padding as placeholder)
+        xyxy_embedding = self.fourier_embedder(boxes.to(dtype))  # B*N*4 --> B*N*C
+
+        # learnable null embedding
+        positive_null = self.null_positive_feature.view(1, 1, -1)
+        xyxy_null = self.null_position_feature.view(1, 1, -1)
+
+        # replace padding with learnable null embedding
+        positive_embeddings = positive_embeddings * \
+            masks + (1 - masks) * positive_null
+        xyxy_embedding = xyxy_embedding * masks + (1 - masks) * xyxy_null
+
+        objs = self.linears(
+            torch.cat([positive_embeddings, xyxy_embedding], dim=-1))
+        assert objs.shape == torch.Size([B, N, self.out_dim])
+        return objs
+
+
+class Gligen(nn.Module):
+    def __init__(self, modules, position_net, key_dim):
+        super().__init__()
+        self.module_list = nn.ModuleList(modules)
+        self.position_net = position_net
+        self.key_dim = key_dim
+        self.max_objs = 30
+        self.current_device = torch.device("cpu")
+
+    def _set_position(self, boxes, masks, positive_embeddings):
+        objs = self.position_net(boxes, masks, positive_embeddings)
+        def func(x, extra_options):
+            key = extra_options["transformer_index"]
+            module = self.module_list[key]
+            return module(x, objs)
+        return func
+
+    def set_position(self, latent_image_shape, position_params, device):
+        batch, c, h, w = latent_image_shape
+        masks = torch.zeros([self.max_objs], device="cpu")
+        boxes = []
+        positive_embeddings = []
+        for p in position_params:
+            x1 = (p[4]) / w
+            y1 = (p[3]) / h
+            x2 = (p[4] + p[2]) / w
+            y2 = (p[3] + p[1]) / h
+            masks[len(boxes)] = 1.0
+            boxes += [torch.tensor((x1, y1, x2, y2)).unsqueeze(0)]
+            positive_embeddings += [p[0]]
+        append_boxes = []
+        append_conds = []
+        if len(boxes) < self.max_objs:
+            append_boxes = [torch.zeros(
+                [self.max_objs - len(boxes), 4], device="cpu")]
+            append_conds = [torch.zeros(
+                [self.max_objs - len(boxes), self.key_dim], device="cpu")]
+
+        box_out = torch.cat(
+            boxes + append_boxes).unsqueeze(0).repeat(batch, 1, 1)
+        masks = masks.unsqueeze(0).repeat(batch, 1)
+        conds = torch.cat(positive_embeddings +
+                          append_conds).unsqueeze(0).repeat(batch, 1, 1)
+        return self._set_position(
+            box_out.to(device),
+            masks.to(device),
+            conds.to(device))
+
+    def set_empty(self, latent_image_shape, device):
+        batch, c, h, w = latent_image_shape
+        masks = torch.zeros([self.max_objs], device="cpu").repeat(batch, 1)
+        box_out = torch.zeros([self.max_objs, 4],
+                              device="cpu").repeat(batch, 1, 1)
+        conds = torch.zeros([self.max_objs, self.key_dim],
+                            device="cpu").repeat(batch, 1, 1)
+        return self._set_position(
+            box_out.to(device),
+            masks.to(device),
+            conds.to(device))
+
+
+def load_gligen(sd):
+    sd_k = sd.keys()
+    output_list = []
+    key_dim = 768
+    for a in ["input_blocks", "middle_block", "output_blocks"]:
+        for b in range(20):
+            k_temp = filter(lambda k: "{}.{}.".format(a, b)
+                            in k and ".fuser." in k, sd_k)
+            k_temp = map(lambda k: (k, k.split(".fuser.")[-1]), k_temp)
+
+            n_sd = {}
+            for k in k_temp:
+                n_sd[k[1]] = sd[k[0]]
+            if len(n_sd) > 0:
+                query_dim = n_sd["linear.weight"].shape[0]
+                key_dim = n_sd["linear.weight"].shape[1]
+
+                if key_dim == 768:  # SD1.x
+                    n_heads = 8
+                    d_head = query_dim // n_heads
+                else:
+                    d_head = 64
+                    n_heads = query_dim // d_head
+
+                gated = GatedSelfAttentionDense(
+                    query_dim, key_dim, n_heads, d_head)
+                gated.load_state_dict(n_sd, strict=False)
+                output_list.append(gated)
+
+    if "position_net.null_positive_feature" in sd_k:
+        in_dim = sd["position_net.null_positive_feature"].shape[0]
+        out_dim = sd["position_net.linears.4.weight"].shape[0]
+
+        class WeightsLoader(torch.nn.Module):
+            pass
+        w = WeightsLoader()
+        w.position_net = PositionNet(in_dim, out_dim)
+        w.load_state_dict(sd, strict=False)
+
+    gligen = Gligen(output_list, w.position_net, key_dim)
+    return gligen

backend/headless/fcbh/k_diffusion/sampling.py

@@ -0,0 +1,752 @@
+import math
+
+from scipy import integrate
+import torch
+from torch import nn
+import torchsde
+from tqdm.auto import trange, tqdm
+
+from . import utils
+
+
+def append_zero(x):
+    return torch.cat([x, x.new_zeros([1])])
+
+
+def get_sigmas_karras(n, sigma_min, sigma_max, rho=7., device='cpu'):
+    """Constructs the noise schedule of Karras et al. (2022)."""
+    ramp = torch.linspace(0, 1, n, device=device)
+    min_inv_rho = sigma_min ** (1 / rho)
+    max_inv_rho = sigma_max ** (1 / rho)
+    sigmas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
+    return append_zero(sigmas).to(device)
+
+
+def get_sigmas_exponential(n, sigma_min, sigma_max, device='cpu'):
+    """Constructs an exponential noise schedule."""
+    sigmas = torch.linspace(math.log(sigma_max), math.log(sigma_min), n, device=device).exp()
+    return append_zero(sigmas)
+
+
+def get_sigmas_polyexponential(n, sigma_min, sigma_max, rho=1., device='cpu'):
+    """Constructs an polynomial in log sigma noise schedule."""
+    ramp = torch.linspace(1, 0, n, device=device) ** rho
+    sigmas = torch.exp(ramp * (math.log(sigma_max) - math.log(sigma_min)) + math.log(sigma_min))
+    return append_zero(sigmas)
+
+
+def get_sigmas_vp(n, beta_d=19.9, beta_min=0.1, eps_s=1e-3, device='cpu'):
+    """Constructs a continuous VP noise schedule."""
+    t = torch.linspace(1, eps_s, n, device=device)
+    sigmas = torch.sqrt(torch.exp(beta_d * t ** 2 / 2 + beta_min * t) - 1)
+    return append_zero(sigmas)
+
+
+def to_d(x, sigma, denoised):
+    """Converts a denoiser output to a Karras ODE derivative."""
+    return (x - denoised) / utils.append_dims(sigma, x.ndim)
+
+
+def get_ancestral_step(sigma_from, sigma_to, eta=1.):
+    """Calculates the noise level (sigma_down) to step down to and the amount
+    of noise to add (sigma_up) when doing an ancestral sampling step."""
+    if not eta:
+        return sigma_to, 0.
+    sigma_up = min(sigma_to, eta * (sigma_to ** 2 * (sigma_from ** 2 - sigma_to ** 2) / sigma_from ** 2) ** 0.5)
+    sigma_down = (sigma_to ** 2 - sigma_up ** 2) ** 0.5
+    return sigma_down, sigma_up
+
+
+def default_noise_sampler(x):
+    return lambda sigma, sigma_next: torch.randn_like(x)
+
+
+class BatchedBrownianTree:
+    """A wrapper around torchsde.BrownianTree that enables batches of entropy."""
+
+    def __init__(self, x, t0, t1, seed=None, **kwargs):
+        self.cpu_tree = True
+        if "cpu" in kwargs:
+            self.cpu_tree = kwargs.pop("cpu")
+        t0, t1, self.sign = self.sort(t0, t1)
+        w0 = kwargs.get('w0', torch.zeros_like(x))
+        if seed is None:
+            seed = torch.randint(0, 2 ** 63 - 1, []).item()
+        self.batched = True
+        try:
+            assert len(seed) == x.shape[0]
+            w0 = w0[0]
+        except TypeError:
+            seed = [seed]
+            self.batched = False
+        if self.cpu_tree:
+            self.trees = [torchsde.BrownianTree(t0.cpu(), w0.cpu(), t1.cpu(), entropy=s, **kwargs) for s in seed]
+        else:
+            self.trees = [torchsde.BrownianTree(t0, w0, t1, entropy=s, **kwargs) for s in seed]
+
+    @staticmethod
+    def sort(a, b):
+        return (a, b, 1) if a < b else (b, a, -1)
+
+    def __call__(self, t0, t1):
+        t0, t1, sign = self.sort(t0, t1)
+        if self.cpu_tree:
+            w = torch.stack([tree(t0.cpu().float(), t1.cpu().float()).to(t0.dtype).to(t0.device) for tree in self.trees]) * (self.sign * sign)
+        else:
+            w = torch.stack([tree(t0, t1) for tree in self.trees]) * (self.sign * sign)
+
+        return w if self.batched else w[0]
+
+
+class BrownianTreeNoiseSampler:
+    """A noise sampler backed by a torchsde.BrownianTree.
+
+    Args:
+        x (Tensor): The tensor whose shape, device and dtype to use to generate
+            random samples.
+        sigma_min (float): The low end of the valid interval.
+        sigma_max (float): The high end of the valid interval.
+        seed (int or List[int]): The random seed. If a list of seeds is
+            supplied instead of a single integer, then the noise sampler will
+            use one BrownianTree per batch item, each with its own seed.
+        transform (callable): A function that maps sigma to the sampler's
+            internal timestep.
+    """
+
+    def __init__(self, x, sigma_min, sigma_max, seed=None, transform=lambda x: x, cpu=False):
+        self.transform = transform
+        t0, t1 = self.transform(torch.as_tensor(sigma_min)), self.transform(torch.as_tensor(sigma_max))
+        self.tree = BatchedBrownianTree(x, t0, t1, seed, cpu=cpu)
+
+    def __call__(self, sigma, sigma_next):
+        t0, t1 = self.transform(torch.as_tensor(sigma)), self.transform(torch.as_tensor(sigma_next))
+        return self.tree(t0, t1) / (t1 - t0).abs().sqrt()
+
+
+@torch.no_grad()
+def sample_euler(model, x, sigmas, extra_args=None, callback=None, disable=None, s_churn=0., s_tmin=0., s_tmax=float('inf'), s_noise=1.):
+    """Implements Algorithm 2 (Euler steps) from Karras et al. (2022)."""
+    extra_args = {} if extra_args is None else extra_args
+    s_in = x.new_ones([x.shape[0]])
+    for i in trange(len(sigmas) - 1, disable=disable):
+        gamma = min(s_churn / (len(sigmas) - 1), 2 ** 0.5 - 1) if s_tmin <= sigmas[i] <= s_tmax else 0.
+        sigma_hat = sigmas[i] * (gamma + 1)
+        if gamma > 0:
+            eps = torch.randn_like(x) * s_noise
+            x = x + eps * (sigma_hat ** 2 - sigmas[i] ** 2) ** 0.5
+        denoised = model(x, sigma_hat * s_in, **extra_args)
+        d = to_d(x, sigma_hat, denoised)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigma_hat, 'denoised': denoised})
+        dt = sigmas[i + 1] - sigma_hat
+        # Euler method
+        x = x + d * dt
+    return x
+
+
+@torch.no_grad()
+def sample_euler_ancestral(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None):
+    """Ancestral sampling with Euler method steps."""
+    extra_args = {} if extra_args is None else extra_args
+    noise_sampler = default_noise_sampler(x) if noise_sampler is None else noise_sampler
+    s_in = x.new_ones([x.shape[0]])
+    for i in trange(len(sigmas) - 1, disable=disable):
+        denoised = model(x, sigmas[i] * s_in, **extra_args)
+        sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1], eta=eta)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
+        d = to_d(x, sigmas[i], denoised)
+        # Euler method
+        dt = sigma_down - sigmas[i]
+        x = x + d * dt
+        if sigmas[i + 1] > 0:
+            x = x + noise_sampler(sigmas[i], sigmas[i + 1]) * s_noise * sigma_up
+    return x
+
+
+@torch.no_grad()
+def sample_heun(model, x, sigmas, extra_args=None, callback=None, disable=None, s_churn=0., s_tmin=0., s_tmax=float('inf'), s_noise=1.):
+    """Implements Algorithm 2 (Heun steps) from Karras et al. (2022)."""
+    extra_args = {} if extra_args is None else extra_args
+    s_in = x.new_ones([x.shape[0]])
+    for i in trange(len(sigmas) - 1, disable=disable):
+        gamma = min(s_churn / (len(sigmas) - 1), 2 ** 0.5 - 1) if s_tmin <= sigmas[i] <= s_tmax else 0.
+        sigma_hat = sigmas[i] * (gamma + 1)
+        if gamma > 0:
+            eps = torch.randn_like(x) * s_noise
+            x = x + eps * (sigma_hat ** 2 - sigmas[i] ** 2) ** 0.5
+        denoised = model(x, sigma_hat * s_in, **extra_args)
+        d = to_d(x, sigma_hat, denoised)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigma_hat, 'denoised': denoised})
+        dt = sigmas[i + 1] - sigma_hat
+        if sigmas[i + 1] == 0:
+            # Euler method
+            x = x + d * dt
+        else:
+            # Heun's method
+            x_2 = x + d * dt
+            denoised_2 = model(x_2, sigmas[i + 1] * s_in, **extra_args)
+            d_2 = to_d(x_2, sigmas[i + 1], denoised_2)
+            d_prime = (d + d_2) / 2
+            x = x + d_prime * dt
+    return x
+
+
+@torch.no_grad()
+def sample_dpm_2(model, x, sigmas, extra_args=None, callback=None, disable=None, s_churn=0., s_tmin=0., s_tmax=float('inf'), s_noise=1.):
+    """A sampler inspired by DPM-Solver-2 and Algorithm 2 from Karras et al. (2022)."""
+    extra_args = {} if extra_args is None else extra_args
+    s_in = x.new_ones([x.shape[0]])
+    for i in trange(len(sigmas) - 1, disable=disable):
+        gamma = min(s_churn / (len(sigmas) - 1), 2 ** 0.5 - 1) if s_tmin <= sigmas[i] <= s_tmax else 0.
+        sigma_hat = sigmas[i] * (gamma + 1)
+        if gamma > 0:
+            eps = torch.randn_like(x) * s_noise
+            x = x + eps * (sigma_hat ** 2 - sigmas[i] ** 2) ** 0.5
+        denoised = model(x, sigma_hat * s_in, **extra_args)
+        d = to_d(x, sigma_hat, denoised)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigma_hat, 'denoised': denoised})
+        if sigmas[i + 1] == 0:
+            # Euler method
+            dt = sigmas[i + 1] - sigma_hat
+            x = x + d * dt
+        else:
+            # DPM-Solver-2
+            sigma_mid = sigma_hat.log().lerp(sigmas[i + 1].log(), 0.5).exp()
+            dt_1 = sigma_mid - sigma_hat
+            dt_2 = sigmas[i + 1] - sigma_hat
+            x_2 = x + d * dt_1
+            denoised_2 = model(x_2, sigma_mid * s_in, **extra_args)
+            d_2 = to_d(x_2, sigma_mid, denoised_2)
+            x = x + d_2 * dt_2
+    return x
+
+
+@torch.no_grad()
+def sample_dpm_2_ancestral(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None):
+    """Ancestral sampling with DPM-Solver second-order steps."""
+    extra_args = {} if extra_args is None else extra_args
+    noise_sampler = default_noise_sampler(x) if noise_sampler is None else noise_sampler
+    s_in = x.new_ones([x.shape[0]])
+    for i in trange(len(sigmas) - 1, disable=disable):
+        denoised = model(x, sigmas[i] * s_in, **extra_args)
+        sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1], eta=eta)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
+        d = to_d(x, sigmas[i], denoised)
+        if sigma_down == 0:
+            # Euler method
+            dt = sigma_down - sigmas[i]
+            x = x + d * dt
+        else:
+            # DPM-Solver-2
+            sigma_mid = sigmas[i].log().lerp(sigma_down.log(), 0.5).exp()
+            dt_1 = sigma_mid - sigmas[i]
+            dt_2 = sigma_down - sigmas[i]
+            x_2 = x + d * dt_1
+            denoised_2 = model(x_2, sigma_mid * s_in, **extra_args)
+            d_2 = to_d(x_2, sigma_mid, denoised_2)
+            x = x + d_2 * dt_2
+            x = x + noise_sampler(sigmas[i], sigmas[i + 1]) * s_noise * sigma_up
+    return x
+
+
+def linear_multistep_coeff(order, t, i, j):
+    if order - 1 > i:
+        raise ValueError(f'Order {order} too high for step {i}')
+    def fn(tau):
+        prod = 1.
+        for k in range(order):
+            if j == k:
+                continue
+            prod *= (tau - t[i - k]) / (t[i - j] - t[i - k])
+        return prod
+    return integrate.quad(fn, t[i], t[i + 1], epsrel=1e-4)[0]
+
+
+@torch.no_grad()
+def sample_lms(model, x, sigmas, extra_args=None, callback=None, disable=None, order=4):
+    extra_args = {} if extra_args is None else extra_args
+    s_in = x.new_ones([x.shape[0]])
+    sigmas_cpu = sigmas.detach().cpu().numpy()
+    ds = []
+    for i in trange(len(sigmas) - 1, disable=disable):
+        denoised = model(x, sigmas[i] * s_in, **extra_args)
+        d = to_d(x, sigmas[i], denoised)
+        ds.append(d)
+        if len(ds) > order:
+            ds.pop(0)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
+        cur_order = min(i + 1, order)
+        coeffs = [linear_multistep_coeff(cur_order, sigmas_cpu, i, j) for j in range(cur_order)]
+        x = x + sum(coeff * d for coeff, d in zip(coeffs, reversed(ds)))
+    return x
+
+
+class PIDStepSizeController:
+    """A PID controller for ODE adaptive step size control."""
+    def __init__(self, h, pcoeff, icoeff, dcoeff, order=1, accept_safety=0.81, eps=1e-8):
+        self.h = h
+        self.b1 = (pcoeff + icoeff + dcoeff) / order
+        self.b2 = -(pcoeff + 2 * dcoeff) / order
+        self.b3 = dcoeff / order
+        self.accept_safety = accept_safety
+        self.eps = eps
+        self.errs = []
+
+    def limiter(self, x):
+        return 1 + math.atan(x - 1)
+
+    def propose_step(self, error):
+        inv_error = 1 / (float(error) + self.eps)
+        if not self.errs:
+            self.errs = [inv_error, inv_error, inv_error]
+        self.errs[0] = inv_error
+        factor = self.errs[0] ** self.b1 * self.errs[1] ** self.b2 * self.errs[2] ** self.b3
+        factor = self.limiter(factor)
+        accept = factor >= self.accept_safety
+        if accept:
+            self.errs[2] = self.errs[1]
+            self.errs[1] = self.errs[0]
+        self.h *= factor
+        return accept
+
+
+class DPMSolver(nn.Module):
+    """DPM-Solver. See https://arxiv.org/abs/2206.00927."""
+
+    def __init__(self, model, extra_args=None, eps_callback=None, info_callback=None):
+        super().__init__()
+        self.model = model
+        self.extra_args = {} if extra_args is None else extra_args
+        self.eps_callback = eps_callback
+        self.info_callback = info_callback
+
+    def t(self, sigma):
+        return -sigma.log()
+
+    def sigma(self, t):
+        return t.neg().exp()
+
+    def eps(self, eps_cache, key, x, t, *args, **kwargs):
+        if key in eps_cache:
+            return eps_cache[key], eps_cache
+        sigma = self.sigma(t) * x.new_ones([x.shape[0]])
+        eps = (x - self.model(x, sigma, *args, **self.extra_args, **kwargs)) / self.sigma(t)
+        if self.eps_callback is not None:
+            self.eps_callback()
+        return eps, {key: eps, **eps_cache}
+
+    def dpm_solver_1_step(self, x, t, t_next, eps_cache=None):
+        eps_cache = {} if eps_cache is None else eps_cache
+        h = t_next - t
+        eps, eps_cache = self.eps(eps_cache, 'eps', x, t)
+        x_1 = x - self.sigma(t_next) * h.expm1() * eps
+        return x_1, eps_cache
+
+    def dpm_solver_2_step(self, x, t, t_next, r1=1 / 2, eps_cache=None):
+        eps_cache = {} if eps_cache is None else eps_cache
+        h = t_next - t
+        eps, eps_cache = self.eps(eps_cache, 'eps', x, t)
+        s1 = t + r1 * h
+        u1 = x - self.sigma(s1) * (r1 * h).expm1() * eps
+        eps_r1, eps_cache = self.eps(eps_cache, 'eps_r1', u1, s1)
+        x_2 = x - self.sigma(t_next) * h.expm1() * eps - self.sigma(t_next) / (2 * r1) * h.expm1() * (eps_r1 - eps)
+        return x_2, eps_cache
+
+    def dpm_solver_3_step(self, x, t, t_next, r1=1 / 3, r2=2 / 3, eps_cache=None):
+        eps_cache = {} if eps_cache is None else eps_cache
+        h = t_next - t
+        eps, eps_cache = self.eps(eps_cache, 'eps', x, t)
+        s1 = t + r1 * h
+        s2 = t + r2 * h
+        u1 = x - self.sigma(s1) * (r1 * h).expm1() * eps
+        eps_r1, eps_cache = self.eps(eps_cache, 'eps_r1', u1, s1)
+        u2 = x - self.sigma(s2) * (r2 * h).expm1() * eps - self.sigma(s2) * (r2 / r1) * ((r2 * h).expm1() / (r2 * h) - 1) * (eps_r1 - eps)
+        eps_r2, eps_cache = self.eps(eps_cache, 'eps_r2', u2, s2)
+        x_3 = x - self.sigma(t_next) * h.expm1() * eps - self.sigma(t_next) / r2 * (h.expm1() / h - 1) * (eps_r2 - eps)
+        return x_3, eps_cache
+
+    def dpm_solver_fast(self, x, t_start, t_end, nfe, eta=0., s_noise=1., noise_sampler=None):
+        noise_sampler = default_noise_sampler(x) if noise_sampler is None else noise_sampler
+        if not t_end > t_start and eta:
+            raise ValueError('eta must be 0 for reverse sampling')
+
+        m = math.floor(nfe / 3) + 1
+        ts = torch.linspace(t_start, t_end, m + 1, device=x.device)
+
+        if nfe % 3 == 0:
+            orders = [3] * (m - 2) + [2, 1]
+        else:
+            orders = [3] * (m - 1) + [nfe % 3]
+
+        for i in range(len(orders)):
+            eps_cache = {}
+            t, t_next = ts[i], ts[i + 1]
+            if eta:
+                sd, su = get_ancestral_step(self.sigma(t), self.sigma(t_next), eta)
+                t_next_ = torch.minimum(t_end, self.t(sd))
+                su = (self.sigma(t_next) ** 2 - self.sigma(t_next_) ** 2) ** 0.5
+            else:
+                t_next_, su = t_next, 0.
+
+            eps, eps_cache = self.eps(eps_cache, 'eps', x, t)
+            denoised = x - self.sigma(t) * eps
+            if self.info_callback is not None:
+                self.info_callback({'x': x, 'i': i, 't': ts[i], 't_up': t, 'denoised': denoised})
+
+            if orders[i] == 1:
+                x, eps_cache = self.dpm_solver_1_step(x, t, t_next_, eps_cache=eps_cache)
+            elif orders[i] == 2:
+                x, eps_cache = self.dpm_solver_2_step(x, t, t_next_, eps_cache=eps_cache)
+            else:
+                x, eps_cache = self.dpm_solver_3_step(x, t, t_next_, eps_cache=eps_cache)
+
+            x = x + su * s_noise * noise_sampler(self.sigma(t), self.sigma(t_next))
+
+        return x
+
+    def dpm_solver_adaptive(self, x, t_start, t_end, order=3, rtol=0.05, atol=0.0078, h_init=0.05, pcoeff=0., icoeff=1., dcoeff=0., accept_safety=0.81, eta=0., s_noise=1., noise_sampler=None):
+        noise_sampler = default_noise_sampler(x) if noise_sampler is None else noise_sampler
+        if order not in {2, 3}:
+            raise ValueError('order should be 2 or 3')
+        forward = t_end > t_start
+        if not forward and eta:
+            raise ValueError('eta must be 0 for reverse sampling')
+        h_init = abs(h_init) * (1 if forward else -1)
+        atol = torch.tensor(atol)
+        rtol = torch.tensor(rtol)
+        s = t_start
+        x_prev = x
+        accept = True
+        pid = PIDStepSizeController(h_init, pcoeff, icoeff, dcoeff, 1.5 if eta else order, accept_safety)
+        info = {'steps': 0, 'nfe': 0, 'n_accept': 0, 'n_reject': 0}
+
+        while s < t_end - 1e-5 if forward else s > t_end + 1e-5:
+            eps_cache = {}
+            t = torch.minimum(t_end, s + pid.h) if forward else torch.maximum(t_end, s + pid.h)
+            if eta:
+                sd, su = get_ancestral_step(self.sigma(s), self.sigma(t), eta)
+                t_ = torch.minimum(t_end, self.t(sd))
+                su = (self.sigma(t) ** 2 - self.sigma(t_) ** 2) ** 0.5
+            else:
+                t_, su = t, 0.
+
+            eps, eps_cache = self.eps(eps_cache, 'eps', x, s)
+            denoised = x - self.sigma(s) * eps
+
+            if order == 2:
+                x_low, eps_cache = self.dpm_solver_1_step(x, s, t_, eps_cache=eps_cache)
+                x_high, eps_cache = self.dpm_solver_2_step(x, s, t_, eps_cache=eps_cache)
+            else:
+                x_low, eps_cache = self.dpm_solver_2_step(x, s, t_, r1=1 / 3, eps_cache=eps_cache)
+                x_high, eps_cache = self.dpm_solver_3_step(x, s, t_, eps_cache=eps_cache)
+            delta = torch.maximum(atol, rtol * torch.maximum(x_low.abs(), x_prev.abs()))
+            error = torch.linalg.norm((x_low - x_high) / delta) / x.numel() ** 0.5
+            accept = pid.propose_step(error)
+            if accept:
+                x_prev = x_low
+                x = x_high + su * s_noise * noise_sampler(self.sigma(s), self.sigma(t))
+                s = t
+                info['n_accept'] += 1
+            else:
+                info['n_reject'] += 1
+            info['nfe'] += order
+            info['steps'] += 1
+
+            if self.info_callback is not None:
+                self.info_callback({'x': x, 'i': info['steps'] - 1, 't': s, 't_up': s, 'denoised': denoised, 'error': error, 'h': pid.h, **info})
+
+        return x, info
+
+
+@torch.no_grad()
+def sample_dpm_fast(model, x, sigma_min, sigma_max, n, extra_args=None, callback=None, disable=None, eta=0., s_noise=1., noise_sampler=None):
+    """DPM-Solver-Fast (fixed step size). See https://arxiv.org/abs/2206.00927."""
+    if sigma_min <= 0 or sigma_max <= 0:
+        raise ValueError('sigma_min and sigma_max must not be 0')
+    with tqdm(total=n, disable=disable) as pbar:
+        dpm_solver = DPMSolver(model, extra_args, eps_callback=pbar.update)
+        if callback is not None:
+            dpm_solver.info_callback = lambda info: callback({'sigma': dpm_solver.sigma(info['t']), 'sigma_hat': dpm_solver.sigma(info['t_up']), **info})
+        return dpm_solver.dpm_solver_fast(x, dpm_solver.t(torch.tensor(sigma_max)), dpm_solver.t(torch.tensor(sigma_min)), n, eta, s_noise, noise_sampler)
+
+
+@torch.no_grad()
+def sample_dpm_adaptive(model, x, sigma_min, sigma_max, extra_args=None, callback=None, disable=None, order=3, rtol=0.05, atol=0.0078, h_init=0.05, pcoeff=0., icoeff=1., dcoeff=0., accept_safety=0.81, eta=0., s_noise=1., noise_sampler=None, return_info=False):
+    """DPM-Solver-12 and 23 (adaptive step size). See https://arxiv.org/abs/2206.00927."""
+    if sigma_min <= 0 or sigma_max <= 0:
+        raise ValueError('sigma_min and sigma_max must not be 0')
+    with tqdm(disable=disable) as pbar:
+        dpm_solver = DPMSolver(model, extra_args, eps_callback=pbar.update)
+        if callback is not None:
+            dpm_solver.info_callback = lambda info: callback({'sigma': dpm_solver.sigma(info['t']), 'sigma_hat': dpm_solver.sigma(info['t_up']), **info})
+        x, info = dpm_solver.dpm_solver_adaptive(x, dpm_solver.t(torch.tensor(sigma_max)), dpm_solver.t(torch.tensor(sigma_min)), order, rtol, atol, h_init, pcoeff, icoeff, dcoeff, accept_safety, eta, s_noise, noise_sampler)
+    if return_info:
+        return x, info
+    return x
+
+
+@torch.no_grad()
+def sample_dpmpp_2s_ancestral(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None):
+    """Ancestral sampling with DPM-Solver++(2S) second-order steps."""
+    extra_args = {} if extra_args is None else extra_args
+    noise_sampler = default_noise_sampler(x) if noise_sampler is None else noise_sampler
+    s_in = x.new_ones([x.shape[0]])
+    sigma_fn = lambda t: t.neg().exp()
+    t_fn = lambda sigma: sigma.log().neg()
+
+    for i in trange(len(sigmas) - 1, disable=disable):
+        denoised = model(x, sigmas[i] * s_in, **extra_args)
+        sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1], eta=eta)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
+        if sigma_down == 0:
+            # Euler method
+            d = to_d(x, sigmas[i], denoised)
+            dt = sigma_down - sigmas[i]
+            x = x + d * dt
+        else:
+            # DPM-Solver++(2S)
+            t, t_next = t_fn(sigmas[i]), t_fn(sigma_down)
+            r = 1 / 2
+            h = t_next - t
+            s = t + r * h
+            x_2 = (sigma_fn(s) / sigma_fn(t)) * x - (-h * r).expm1() * denoised
+            denoised_2 = model(x_2, sigma_fn(s) * s_in, **extra_args)
+            x = (sigma_fn(t_next) / sigma_fn(t)) * x - (-h).expm1() * denoised_2
+        # Noise addition
+        if sigmas[i + 1] > 0:
+            x = x + noise_sampler(sigmas[i], sigmas[i + 1]) * s_noise * sigma_up
+    return x
+
+
+@torch.no_grad()
+def sample_dpmpp_sde(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None, r=1 / 2):
+    """DPM-Solver++ (stochastic)."""
+    sigma_min, sigma_max = sigmas[sigmas > 0].min(), sigmas.max()
+    seed = extra_args.get("seed", None)
+    noise_sampler = BrownianTreeNoiseSampler(x, sigma_min, sigma_max, seed=seed, cpu=True) if noise_sampler is None else noise_sampler
+    extra_args = {} if extra_args is None else extra_args
+    s_in = x.new_ones([x.shape[0]])
+    sigma_fn = lambda t: t.neg().exp()
+    t_fn = lambda sigma: sigma.log().neg()
+
+    for i in trange(len(sigmas) - 1, disable=disable):
+        denoised = model(x, sigmas[i] * s_in, **extra_args)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
+        if sigmas[i + 1] == 0:
+            # Euler method
+            d = to_d(x, sigmas[i], denoised)
+            dt = sigmas[i + 1] - sigmas[i]
+            x = x + d * dt
+        else:
+            # DPM-Solver++
+            t, t_next = t_fn(sigmas[i]), t_fn(sigmas[i + 1])
+            h = t_next - t
+            s = t + h * r
+            fac = 1 / (2 * r)
+
+            # Step 1
+            sd, su = get_ancestral_step(sigma_fn(t), sigma_fn(s), eta)
+            s_ = t_fn(sd)
+            x_2 = (sigma_fn(s_) / sigma_fn(t)) * x - (t - s_).expm1() * denoised
+            x_2 = x_2 + noise_sampler(sigma_fn(t), sigma_fn(s)) * s_noise * su
+            denoised_2 = model(x_2, sigma_fn(s) * s_in, **extra_args)
+
+            # Step 2
+            sd, su = get_ancestral_step(sigma_fn(t), sigma_fn(t_next), eta)
+            t_next_ = t_fn(sd)
+            denoised_d = (1 - fac) * denoised + fac * denoised_2
+            x = (sigma_fn(t_next_) / sigma_fn(t)) * x - (t - t_next_).expm1() * denoised_d
+            x = x + noise_sampler(sigma_fn(t), sigma_fn(t_next)) * s_noise * su
+    return x
+
+
+@torch.no_grad()
+def sample_dpmpp_2m(model, x, sigmas, extra_args=None, callback=None, disable=None):
+    """DPM-Solver++(2M)."""
+    extra_args = {} if extra_args is None else extra_args
+    s_in = x.new_ones([x.shape[0]])
+    sigma_fn = lambda t: t.neg().exp()
+    t_fn = lambda sigma: sigma.log().neg()
+    old_denoised = None
+
+    for i in trange(len(sigmas) - 1, disable=disable):
+        denoised = model(x, sigmas[i] * s_in, **extra_args)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
+        t, t_next = t_fn(sigmas[i]), t_fn(sigmas[i + 1])
+        h = t_next - t
+        if old_denoised is None or sigmas[i + 1] == 0:
+            x = (sigma_fn(t_next) / sigma_fn(t)) * x - (-h).expm1() * denoised
+        else:
+            h_last = t - t_fn(sigmas[i - 1])
+            r = h_last / h
+            denoised_d = (1 + 1 / (2 * r)) * denoised - (1 / (2 * r)) * old_denoised
+            x = (sigma_fn(t_next) / sigma_fn(t)) * x - (-h).expm1() * denoised_d
+        old_denoised = denoised
+    return x
+
+@torch.no_grad()
+def sample_dpmpp_2m_sde(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None, solver_type='midpoint'):
+    """DPM-Solver++(2M) SDE."""
+
+    if solver_type not in {'heun', 'midpoint'}:
+        raise ValueError('solver_type must be \'heun\' or \'midpoint\'')
+
+    seed = extra_args.get("seed", None)
+    sigma_min, sigma_max = sigmas[sigmas > 0].min(), sigmas.max()
+    noise_sampler = BrownianTreeNoiseSampler(x, sigma_min, sigma_max, seed=seed, cpu=True) if noise_sampler is None else noise_sampler
+    extra_args = {} if extra_args is None else extra_args
+    s_in = x.new_ones([x.shape[0]])
+
+    old_denoised = None
+    h_last = None
+    h = None
+
+    for i in trange(len(sigmas) - 1, disable=disable):
+        denoised = model(x, sigmas[i] * s_in, **extra_args)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
+        if sigmas[i + 1] == 0:
+            # Denoising step
+            x = denoised
+        else:
+            # DPM-Solver++(2M) SDE
+            t, s = -sigmas[i].log(), -sigmas[i + 1].log()
+            h = s - t
+            eta_h = eta * h
+
+            x = sigmas[i + 1] / sigmas[i] * (-eta_h).exp() * x + (-h - eta_h).expm1().neg() * denoised
+
+            if old_denoised is not None:
+                r = h_last / h
+                if solver_type == 'heun':
+                    x = x + ((-h - eta_h).expm1().neg() / (-h - eta_h) + 1) * (1 / r) * (denoised - old_denoised)
+                elif solver_type == 'midpoint':
+                    x = x + 0.5 * (-h - eta_h).expm1().neg() * (1 / r) * (denoised - old_denoised)
+
+            if eta:
+                x = x + noise_sampler(sigmas[i], sigmas[i + 1]) * sigmas[i + 1] * (-2 * eta_h).expm1().neg().sqrt() * s_noise
+
+        old_denoised = denoised
+        h_last = h
+    return x
+
+@torch.no_grad()
+def sample_dpmpp_3m_sde(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None):
+    """DPM-Solver++(3M) SDE."""
+
+    seed = extra_args.get("seed", None)
+    sigma_min, sigma_max = sigmas[sigmas > 0].min(), sigmas.max()
+    noise_sampler = BrownianTreeNoiseSampler(x, sigma_min, sigma_max, seed=seed, cpu=True) if noise_sampler is None else noise_sampler
+    extra_args = {} if extra_args is None else extra_args
+    s_in = x.new_ones([x.shape[0]])
+
+    denoised_1, denoised_2 = None, None
+    h, h_1, h_2 = None, None, None
+
+    for i in trange(len(sigmas) - 1, disable=disable):
+        denoised = model(x, sigmas[i] * s_in, **extra_args)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
+        if sigmas[i + 1] == 0:
+            # Denoising step
+            x = denoised
+        else:
+            t, s = -sigmas[i].log(), -sigmas[i + 1].log()
+            h = s - t
+            h_eta = h * (eta + 1)
+
+            x = torch.exp(-h_eta) * x + (-h_eta).expm1().neg() * denoised
+
+            if h_2 is not None:
+                r0 = h_1 / h
+                r1 = h_2 / h
+                d1_0 = (denoised - denoised_1) / r0
+                d1_1 = (denoised_1 - denoised_2) / r1
+                d1 = d1_0 + (d1_0 - d1_1) * r0 / (r0 + r1)
+                d2 = (d1_0 - d1_1) / (r0 + r1)
+                phi_2 = h_eta.neg().expm1() / h_eta + 1
+                phi_3 = phi_2 / h_eta - 0.5
+                x = x + phi_2 * d1 - phi_3 * d2
+            elif h_1 is not None:
+                r = h_1 / h
+                d = (denoised - denoised_1) / r
+                phi_2 = h_eta.neg().expm1() / h_eta + 1
+                x = x + phi_2 * d
+
+            if eta:
+                x = x + noise_sampler(sigmas[i], sigmas[i + 1]) * sigmas[i + 1] * (-2 * h * eta).expm1().neg().sqrt() * s_noise
+
+        denoised_1, denoised_2 = denoised, denoised_1
+        h_1, h_2 = h, h_1
+    return x
+
+@torch.no_grad()
+def sample_dpmpp_3m_sde_gpu(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None):
+    sigma_min, sigma_max = sigmas[sigmas > 0].min(), sigmas.max()
+    noise_sampler = BrownianTreeNoiseSampler(x, sigma_min, sigma_max, seed=extra_args.get("seed", None), cpu=False) if noise_sampler is None else noise_sampler
+    return sample_dpmpp_3m_sde(model, x, sigmas, extra_args=extra_args, callback=callback, disable=disable, eta=eta, s_noise=s_noise, noise_sampler=noise_sampler)
+
+@torch.no_grad()
+def sample_dpmpp_2m_sde_gpu(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None, solver_type='midpoint'):
+    sigma_min, sigma_max = sigmas[sigmas > 0].min(), sigmas.max()
+    noise_sampler = BrownianTreeNoiseSampler(x, sigma_min, sigma_max, seed=extra_args.get("seed", None), cpu=False) if noise_sampler is None else noise_sampler
+    return sample_dpmpp_2m_sde(model, x, sigmas, extra_args=extra_args, callback=callback, disable=disable, eta=eta, s_noise=s_noise, noise_sampler=noise_sampler, solver_type=solver_type)
+
+@torch.no_grad()
+def sample_dpmpp_sde_gpu(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None, r=1 / 2):
+    sigma_min, sigma_max = sigmas[sigmas > 0].min(), sigmas.max()
+    noise_sampler = BrownianTreeNoiseSampler(x, sigma_min, sigma_max, seed=extra_args.get("seed", None), cpu=False) if noise_sampler is None else noise_sampler
+    return sample_dpmpp_sde(model, x, sigmas, extra_args=extra_args, callback=callback, disable=disable, eta=eta, s_noise=s_noise, noise_sampler=noise_sampler, r=r)
+
+
+def DDPMSampler_step(x, sigma, sigma_prev, noise, noise_sampler):
+    alpha_cumprod = 1 / ((sigma * sigma) + 1)
+    alpha_cumprod_prev = 1 / ((sigma_prev * sigma_prev) + 1)
+    alpha = (alpha_cumprod / alpha_cumprod_prev)
+
+    mu = (1.0 / alpha).sqrt() * (x - (1 - alpha) * noise / (1 - alpha_cumprod).sqrt())
+    if sigma_prev > 0:
+        mu += ((1 - alpha) * (1. - alpha_cumprod_prev) / (1. - alpha_cumprod)).sqrt() * noise_sampler(sigma, sigma_prev)
+    return mu
+
+def generic_step_sampler(model, x, sigmas, extra_args=None, callback=None, disable=None, noise_sampler=None, step_function=None):
+    extra_args = {} if extra_args is None else extra_args
+    noise_sampler = default_noise_sampler(x) if noise_sampler is None else noise_sampler
+    s_in = x.new_ones([x.shape[0]])
+
+    for i in trange(len(sigmas) - 1, disable=disable):
+        denoised = model(x, sigmas[i] * s_in, **extra_args)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
+        x = step_function(x / torch.sqrt(1.0 + sigmas[i] ** 2.0), sigmas[i], sigmas[i + 1], (x - denoised) / sigmas[i], noise_sampler)
+        if sigmas[i + 1] != 0:
+            x *= torch.sqrt(1.0 + sigmas[i + 1] ** 2.0)
+    return x
+
+
+@torch.no_grad()
+def sample_ddpm(model, x, sigmas, extra_args=None, callback=None, disable=None, noise_sampler=None):
+    return generic_step_sampler(model, x, sigmas, extra_args, callback, disable, noise_sampler, DDPMSampler_step)
+
+@torch.no_grad()
+def sample_lcm(model, x, sigmas, extra_args=None, callback=None, disable=None, noise_sampler=None):
+    extra_args = {} if extra_args is None else extra_args
+    noise_sampler = default_noise_sampler(x) if noise_sampler is None else noise_sampler
+    s_in = x.new_ones([x.shape[0]])
+    for i in trange(len(sigmas) - 1, disable=disable):
+        denoised = model(x, sigmas[i] * s_in, **extra_args)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
+
+        x = denoised
+        if sigmas[i + 1] > 0:
+            x += sigmas[i + 1] * noise_sampler(sigmas[i], sigmas[i + 1])
+    return x

backend/headless/fcbh/k_diffusion/utils.py

@@ -0,0 +1,313 @@
+from contextlib import contextmanager
+import hashlib
+import math
+from pathlib import Path
+import shutil
+import urllib
+import warnings
+
+from PIL import Image
+import torch
+from torch import nn, optim
+from torch.utils import data
+
+
+def hf_datasets_augs_helper(examples, transform, image_key, mode='RGB'):
+    """Apply passed in transforms for HuggingFace Datasets."""
+    images = [transform(image.convert(mode)) for image in examples[image_key]]
+    return {image_key: images}
+
+
+def append_dims(x, target_dims):
+    """Appends dimensions to the end of a tensor until it has target_dims dimensions."""
+    dims_to_append = target_dims - x.ndim
+    if dims_to_append < 0:
+        raise ValueError(f'input has {x.ndim} dims but target_dims is {target_dims}, which is less')
+    expanded = x[(...,) + (None,) * dims_to_append]
+    # MPS will get inf values if it tries to index into the new axes, but detaching fixes this.
+    # https://github.com/pytorch/pytorch/issues/84364
+    return expanded.detach().clone() if expanded.device.type == 'mps' else expanded
+
+
+def n_params(module):
+    """Returns the number of trainable parameters in a module."""
+    return sum(p.numel() for p in module.parameters())
+
+
+def download_file(path, url, digest=None):
+    """Downloads a file if it does not exist, optionally checking its SHA-256 hash."""
+    path = Path(path)
+    path.parent.mkdir(parents=True, exist_ok=True)
+    if not path.exists():
+        with urllib.request.urlopen(url) as response, open(path, 'wb') as f:
+            shutil.copyfileobj(response, f)
+    if digest is not None:
+        file_digest = hashlib.sha256(open(path, 'rb').read()).hexdigest()
+        if digest != file_digest:
+            raise OSError(f'hash of {path} (url: {url}) failed to validate')
+    return path
+
+
+@contextmanager
+def train_mode(model, mode=True):
+    """A context manager that places a model into training mode and restores
+    the previous mode on exit."""
+    modes = [module.training for module in model.modules()]
+    try:
+        yield model.train(mode)
+    finally:
+        for i, module in enumerate(model.modules()):
+            module.training = modes[i]
+
+
+def eval_mode(model):
+    """A context manager that places a model into evaluation mode and restores
+    the previous mode on exit."""
+    return train_mode(model, False)
+
+
+@torch.no_grad()
+def ema_update(model, averaged_model, decay):
+    """Incorporates updated model parameters into an exponential moving averaged
+    version of a model. It should be called after each optimizer step."""
+    model_params = dict(model.named_parameters())
+    averaged_params = dict(averaged_model.named_parameters())
+    assert model_params.keys() == averaged_params.keys()
+
+    for name, param in model_params.items():
+        averaged_params[name].mul_(decay).add_(param, alpha=1 - decay)
+
+    model_buffers = dict(model.named_buffers())
+    averaged_buffers = dict(averaged_model.named_buffers())
+    assert model_buffers.keys() == averaged_buffers.keys()
+
+    for name, buf in model_buffers.items():
+        averaged_buffers[name].copy_(buf)
+
+
+class EMAWarmup:
+    """Implements an EMA warmup using an inverse decay schedule.
+    If inv_gamma=1 and power=1, implements a simple average. inv_gamma=1, power=2/3 are
+    good values for models you plan to train for a million or more steps (reaches decay
+    factor 0.999 at 31.6K steps, 0.9999 at 1M steps), inv_gamma=1, power=3/4 for models
+    you plan to train for less (reaches decay factor 0.999 at 10K steps, 0.9999 at
+    215.4k steps).
+    Args:
+        inv_gamma (float): Inverse multiplicative factor of EMA warmup. Default: 1.
+        power (float): Exponential factor of EMA warmup. Default: 1.
+        min_value (float): The minimum EMA decay rate. Default: 0.
+        max_value (float): The maximum EMA decay rate. Default: 1.
+        start_at (int): The epoch to start averaging at. Default: 0.
+        last_epoch (int): The index of last epoch. Default: 0.
+    """
+
+    def __init__(self, inv_gamma=1., power=1., min_value=0., max_value=1., start_at=0,
+                 last_epoch=0):
+        self.inv_gamma = inv_gamma
+        self.power = power
+        self.min_value = min_value
+        self.max_value = max_value
+        self.start_at = start_at
+        self.last_epoch = last_epoch
+
+    def state_dict(self):
+        """Returns the state of the class as a :class:`dict`."""
+        return dict(self.__dict__.items())
+
+    def load_state_dict(self, state_dict):
+        """Loads the class's state.
+        Args:
+            state_dict (dict): scaler state. Should be an object returned
+                from a call to :meth:`state_dict`.
+        """
+        self.__dict__.update(state_dict)
+
+    def get_value(self):
+        """Gets the current EMA decay rate."""
+        epoch = max(0, self.last_epoch - self.start_at)
+        value = 1 - (1 + epoch / self.inv_gamma) ** -self.power
+        return 0. if epoch < 0 else min(self.max_value, max(self.min_value, value))
+
+    def step(self):
+        """Updates the step count."""
+        self.last_epoch += 1
+
+
+class InverseLR(optim.lr_scheduler._LRScheduler):
+    """Implements an inverse decay learning rate schedule with an optional exponential
+    warmup. When last_epoch=-1, sets initial lr as lr.
+    inv_gamma is the number of steps/epochs required for the learning rate to decay to
+    (1 / 2)**power of its original value.
+    Args:
+        optimizer (Optimizer): Wrapped optimizer.
+        inv_gamma (float): Inverse multiplicative factor of learning rate decay. Default: 1.
+        power (float): Exponential factor of learning rate decay. Default: 1.
+        warmup (float): Exponential warmup factor (0 <= warmup < 1, 0 to disable)
+            Default: 0.
+        min_lr (float): The minimum learning rate. Default: 0.
+        last_epoch (int): The index of last epoch. Default: -1.
+        verbose (bool): If ``True``, prints a message to stdout for
+            each update. Default: ``False``.
+    """
+
+    def __init__(self, optimizer, inv_gamma=1., power=1., warmup=0., min_lr=0.,
+                 last_epoch=-1, verbose=False):
+        self.inv_gamma = inv_gamma
+        self.power = power
+        if not 0. <= warmup < 1:
+            raise ValueError('Invalid value for warmup')
+        self.warmup = warmup
+        self.min_lr = min_lr
+        super().__init__(optimizer, last_epoch, verbose)
+
+    def get_lr(self):
+        if not self._get_lr_called_within_step:
+            warnings.warn("To get the last learning rate computed by the scheduler, "
+                          "please use `get_last_lr()`.")
+
+        return self._get_closed_form_lr()
+
+    def _get_closed_form_lr(self):
+        warmup = 1 - self.warmup ** (self.last_epoch + 1)
+        lr_mult = (1 + self.last_epoch / self.inv_gamma) ** -self.power
+        return [warmup * max(self.min_lr, base_lr * lr_mult)
+                for base_lr in self.base_lrs]
+
+
+class ExponentialLR(optim.lr_scheduler._LRScheduler):
+    """Implements an exponential learning rate schedule with an optional exponential
+    warmup. When last_epoch=-1, sets initial lr as lr. Decays the learning rate
+    continuously by decay (default 0.5) every num_steps steps.
+    Args:
+        optimizer (Optimizer): Wrapped optimizer.
+        num_steps (float): The number of steps to decay the learning rate by decay in.
+        decay (float): The factor by which to decay the learning rate every num_steps
+            steps. Default: 0.5.
+        warmup (float): Exponential warmup factor (0 <= warmup < 1, 0 to disable)
+            Default: 0.
+        min_lr (float): The minimum learning rate. Default: 0.
+        last_epoch (int): The index of last epoch. Default: -1.
+        verbose (bool): If ``True``, prints a message to stdout for
+            each update. Default: ``False``.
+    """
+
+    def __init__(self, optimizer, num_steps, decay=0.5, warmup=0., min_lr=0.,
+                 last_epoch=-1, verbose=False):
+        self.num_steps = num_steps
+        self.decay = decay
+        if not 0. <= warmup < 1:
+            raise ValueError('Invalid value for warmup')
+        self.warmup = warmup
+        self.min_lr = min_lr
+        super().__init__(optimizer, last_epoch, verbose)
+
+    def get_lr(self):
+        if not self._get_lr_called_within_step:
+            warnings.warn("To get the last learning rate computed by the scheduler, "
+                          "please use `get_last_lr()`.")
+
+        return self._get_closed_form_lr()
+
+    def _get_closed_form_lr(self):
+        warmup = 1 - self.warmup ** (self.last_epoch + 1)
+        lr_mult = (self.decay ** (1 / self.num_steps)) ** self.last_epoch
+        return [warmup * max(self.min_lr, base_lr * lr_mult)
+                for base_lr in self.base_lrs]
+
+
+def rand_log_normal(shape, loc=0., scale=1., device='cpu', dtype=torch.float32):
+    """Draws samples from an lognormal distribution."""
+    return (torch.randn(shape, device=device, dtype=dtype) * scale + loc).exp()
+
+
+def rand_log_logistic(shape, loc=0., scale=1., min_value=0., max_value=float('inf'), device='cpu', dtype=torch.float32):
+    """Draws samples from an optionally truncated log-logistic distribution."""
+    min_value = torch.as_tensor(min_value, device=device, dtype=torch.float64)
+    max_value = torch.as_tensor(max_value, device=device, dtype=torch.float64)
+    min_cdf = min_value.log().sub(loc).div(scale).sigmoid()
+    max_cdf = max_value.log().sub(loc).div(scale).sigmoid()
+    u = torch.rand(shape, device=device, dtype=torch.float64) * (max_cdf - min_cdf) + min_cdf
+    return u.logit().mul(scale).add(loc).exp().to(dtype)
+
+
+def rand_log_uniform(shape, min_value, max_value, device='cpu', dtype=torch.float32):
+    """Draws samples from an log-uniform distribution."""
+    min_value = math.log(min_value)
+    max_value = math.log(max_value)
+    return (torch.rand(shape, device=device, dtype=dtype) * (max_value - min_value) + min_value).exp()
+
+
+def rand_v_diffusion(shape, sigma_data=1., min_value=0., max_value=float('inf'), device='cpu', dtype=torch.float32):
+    """Draws samples from a truncated v-diffusion training timestep distribution."""
+    min_cdf = math.atan(min_value / sigma_data) * 2 / math.pi
+    max_cdf = math.atan(max_value / sigma_data) * 2 / math.pi
+    u = torch.rand(shape, device=device, dtype=dtype) * (max_cdf - min_cdf) + min_cdf
+    return torch.tan(u * math.pi / 2) * sigma_data
+
+
+def rand_split_log_normal(shape, loc, scale_1, scale_2, device='cpu', dtype=torch.float32):
+    """Draws samples from a split lognormal distribution."""
+    n = torch.randn(shape, device=device, dtype=dtype).abs()
+    u = torch.rand(shape, device=device, dtype=dtype)
+    n_left = n * -scale_1 + loc
+    n_right = n * scale_2 + loc
+    ratio = scale_1 / (scale_1 + scale_2)
+    return torch.where(u < ratio, n_left, n_right).exp()
+
+
+class FolderOfImages(data.Dataset):
+    """Recursively finds all images in a directory. It does not support
+    classes/targets."""
+
+    IMG_EXTENSIONS = {'.jpg', '.jpeg', '.png', '.ppm', '.bmp', '.pgm', '.tif', '.tiff', '.webp'}
+
+    def __init__(self, root, transform=None):
+        super().__init__()
+        self.root = Path(root)
+        self.transform = nn.Identity() if transform is None else transform
+        self.paths = sorted(path for path in self.root.rglob('*') if path.suffix.lower() in self.IMG_EXTENSIONS)
+
+    def __repr__(self):
+        return f'FolderOfImages(root="{self.root}", len: {len(self)})'
+
+    def __len__(self):
+        return len(self.paths)
+
+    def __getitem__(self, key):
+        path = self.paths[key]
+        with open(path, 'rb') as f:
+            image = Image.open(f).convert('RGB')
+        image = self.transform(image)
+        return image,
+
+
+class CSVLogger:
+    def __init__(self, filename, columns):
+        self.filename = Path(filename)
+        self.columns = columns
+        if self.filename.exists():
+            self.file = open(self.filename, 'a')
+        else:
+            self.file = open(self.filename, 'w')
+            self.write(*self.columns)
+
+    def write(self, *args):
+        print(*args, sep=',', file=self.file, flush=True)
+
+
+@contextmanager
+def tf32_mode(cudnn=None, matmul=None):
+    """A context manager that sets whether TF32 is allowed on cuDNN or matmul."""
+    cudnn_old = torch.backends.cudnn.allow_tf32
+    matmul_old = torch.backends.cuda.matmul.allow_tf32
+    try:
+        if cudnn is not None:
+            torch.backends.cudnn.allow_tf32 = cudnn
+        if matmul is not None:
+            torch.backends.cuda.matmul.allow_tf32 = matmul
+        yield
+    finally:
+        if cudnn is not None:
+            torch.backends.cudnn.allow_tf32 = cudnn_old
+        if matmul is not None:
+            torch.backends.cuda.matmul.allow_tf32 = matmul_old

backend/headless/fcbh/latent_formats.py

@@ -0,0 +1,35 @@
+
+class LatentFormat:
+    scale_factor = 1.0
+    latent_rgb_factors = None
+    taesd_decoder_name = None
+
+    def process_in(self, latent):
+        return latent * self.scale_factor
+
+    def process_out(self, latent):
+        return latent / self.scale_factor
+
+class SD15(LatentFormat):
+    def __init__(self, scale_factor=0.18215):
+        self.scale_factor = scale_factor
+        self.latent_rgb_factors = [
+                    #   R        G        B
+                    [ 0.3512,  0.2297,  0.3227],
+                    [ 0.3250,  0.4974,  0.2350],
+                    [-0.2829,  0.1762,  0.2721],
+                    [-0.2120, -0.2616, -0.7177]
+                ]
+        self.taesd_decoder_name = "taesd_decoder"
+
+class SDXL(LatentFormat):
+    def __init__(self):
+        self.scale_factor = 0.13025
+        self.latent_rgb_factors = [
+                    #   R        G        B
+                    [ 0.3920,  0.4054,  0.4549],
+                    [-0.2634, -0.0196,  0.0653],
+                    [ 0.0568,  0.1687, -0.0755],
+                    [-0.3112, -0.2359, -0.2076]
+                ]
+        self.taesd_decoder_name = "taesdxl_decoder"

backend/headless/fcbh/ldm/models/autoencoder.py

@@ -0,0 +1,227 @@
+import torch
+# import pytorch_lightning as pl
+import torch.nn.functional as F
+from contextlib import contextmanager
+from typing import Any, Dict, List, Optional, Tuple, Union
+
+from fcbh.ldm.modules.distributions.distributions import DiagonalGaussianDistribution
+
+from fcbh.ldm.util import instantiate_from_config
+from fcbh.ldm.modules.ema import LitEma
+
+class DiagonalGaussianRegularizer(torch.nn.Module):
+    def __init__(self, sample: bool = True):
+        super().__init__()
+        self.sample = sample
+
+    def get_trainable_parameters(self) -> Any:
+        yield from ()
+
+    def forward(self, z: torch.Tensor) -> Tuple[torch.Tensor, dict]:
+        log = dict()
+        posterior = DiagonalGaussianDistribution(z)
+        if self.sample:
+            z = posterior.sample()
+        else:
+            z = posterior.mode()
+        kl_loss = posterior.kl()
+        kl_loss = torch.sum(kl_loss) / kl_loss.shape[0]
+        log["kl_loss"] = kl_loss
+        return z, log
+
+
+class AbstractAutoencoder(torch.nn.Module):
+    """
+    This is the base class for all autoencoders, including image autoencoders, image autoencoders with discriminators,
+    unCLIP models, etc. Hence, it is fairly general, and specific features
+    (e.g. discriminator training, encoding, decoding) must be implemented in subclasses.
+    """
+
+    def __init__(
+        self,
+        ema_decay: Union[None, float] = None,
+        monitor: Union[None, str] = None,
+        input_key: str = "jpg",
+        **kwargs,
+    ):
+        super().__init__()
+
+        self.input_key = input_key
+        self.use_ema = ema_decay is not None
+        if monitor is not None:
+            self.monitor = monitor
+
+        if self.use_ema:
+            self.model_ema = LitEma(self, decay=ema_decay)
+            logpy.info(f"Keeping EMAs of {len(list(self.model_ema.buffers()))}.")
+
+    def get_input(self, batch) -> Any:
+        raise NotImplementedError()
+
+    def on_train_batch_end(self, *args, **kwargs):
+        # for EMA computation
+        if self.use_ema:
+            self.model_ema(self)
+
+    @contextmanager
+    def ema_scope(self, context=None):
+        if self.use_ema:
+            self.model_ema.store(self.parameters())
+            self.model_ema.copy_to(self)
+            if context is not None:
+                logpy.info(f"{context}: Switched to EMA weights")
+        try:
+            yield None
+        finally:
+            if self.use_ema:
+                self.model_ema.restore(self.parameters())
+                if context is not None:
+                    logpy.info(f"{context}: Restored training weights")
+
+    def encode(self, *args, **kwargs) -> torch.Tensor:
+        raise NotImplementedError("encode()-method of abstract base class called")
+
+    def decode(self, *args, **kwargs) -> torch.Tensor:
+        raise NotImplementedError("decode()-method of abstract base class called")
+
+    def instantiate_optimizer_from_config(self, params, lr, cfg):
+        logpy.info(f"loading >>> {cfg['target']} <<< optimizer from config")
+        return get_obj_from_str(cfg["target"])(
+            params, lr=lr, **cfg.get("params", dict())
+        )
+
+    def configure_optimizers(self) -> Any:
+        raise NotImplementedError()
+
+
+class AutoencodingEngine(AbstractAutoencoder):
+    """
+    Base class for all image autoencoders that we train, like VQGAN or AutoencoderKL
+    (we also restore them explicitly as special cases for legacy reasons).
+    Regularizations such as KL or VQ are moved to the regularizer class.
+    """
+
+    def __init__(
+        self,
+        *args,
+        encoder_config: Dict,
+        decoder_config: Dict,
+        regularizer_config: Dict,
+        **kwargs,
+    ):
+        super().__init__(*args, **kwargs)
+
+        self.encoder: torch.nn.Module = instantiate_from_config(encoder_config)
+        self.decoder: torch.nn.Module = instantiate_from_config(decoder_config)
+        self.regularization: AbstractRegularizer = instantiate_from_config(
+            regularizer_config
+        )
+
+    def get_last_layer(self):
+        return self.decoder.get_last_layer()
+
+    def encode(
+        self,
+        x: torch.Tensor,
+        return_reg_log: bool = False,
+        unregularized: bool = False,
+    ) -> Union[torch.Tensor, Tuple[torch.Tensor, dict]]:
+        z = self.encoder(x)
+        if unregularized:
+            return z, dict()
+        z, reg_log = self.regularization(z)
+        if return_reg_log:
+            return z, reg_log
+        return z
+
+    def decode(self, z: torch.Tensor, **kwargs) -> torch.Tensor:
+        x = self.decoder(z, **kwargs)
+        return x
+
+    def forward(
+        self, x: torch.Tensor, **additional_decode_kwargs
+    ) -> Tuple[torch.Tensor, torch.Tensor, dict]:
+        z, reg_log = self.encode(x, return_reg_log=True)
+        dec = self.decode(z, **additional_decode_kwargs)
+        return z, dec, reg_log
+
+
+class AutoencodingEngineLegacy(AutoencodingEngine):
+    def __init__(self, embed_dim: int, **kwargs):
+        self.max_batch_size = kwargs.pop("max_batch_size", None)
+        ddconfig = kwargs.pop("ddconfig")
+        super().__init__(
+            encoder_config={
+                "target": "fcbh.ldm.modules.diffusionmodules.model.Encoder",
+                "params": ddconfig,
+            },
+            decoder_config={
+                "target": "fcbh.ldm.modules.diffusionmodules.model.Decoder",
+                "params": ddconfig,
+            },
+            **kwargs,
+        )
+        self.quant_conv = torch.nn.Conv2d(
+            (1 + ddconfig["double_z"]) * ddconfig["z_channels"],
+            (1 + ddconfig["double_z"]) * embed_dim,
+            1,
+        )
+        self.post_quant_conv = torch.nn.Conv2d(embed_dim, ddconfig["z_channels"], 1)
+        self.embed_dim = embed_dim
+
+    def get_autoencoder_params(self) -> list:
+        params = super().get_autoencoder_params()
+        return params
+
+    def encode(
+        self, x: torch.Tensor, return_reg_log: bool = False
+    ) -> Union[torch.Tensor, Tuple[torch.Tensor, dict]]:
+        if self.max_batch_size is None:
+            z = self.encoder(x)
+            z = self.quant_conv(z)
+        else:
+            N = x.shape[0]
+            bs = self.max_batch_size
+            n_batches = int(math.ceil(N / bs))
+            z = list()
+            for i_batch in range(n_batches):
+                z_batch = self.encoder(x[i_batch * bs : (i_batch + 1) * bs])
+                z_batch = self.quant_conv(z_batch)
+                z.append(z_batch)
+            z = torch.cat(z, 0)
+
+        z, reg_log = self.regularization(z)
+        if return_reg_log:
+            return z, reg_log
+        return z
+
+    def decode(self, z: torch.Tensor, **decoder_kwargs) -> torch.Tensor:
+        if self.max_batch_size is None:
+            dec = self.post_quant_conv(z)
+            dec = self.decoder(dec, **decoder_kwargs)
+        else:
+            N = z.shape[0]
+            bs = self.max_batch_size
+            n_batches = int(math.ceil(N / bs))
+            dec = list()
+            for i_batch in range(n_batches):
+                dec_batch = self.post_quant_conv(z[i_batch * bs : (i_batch + 1) * bs])
+                dec_batch = self.decoder(dec_batch, **decoder_kwargs)
+                dec.append(dec_batch)
+            dec = torch.cat(dec, 0)
+
+        return dec
+
+
+class AutoencoderKL(AutoencodingEngineLegacy):
+    def __init__(self, **kwargs):
+        if "lossconfig" in kwargs:
+            kwargs["loss_config"] = kwargs.pop("lossconfig")
+        super().__init__(
+            regularizer_config={
+                "target": (
+                    "fcbh.ldm.models.autoencoder.DiagonalGaussianRegularizer"
+                )
+            },
+            **kwargs,
+        )

backend/headless/fcbh/ldm/modules/attention.py

@@ -0,0 +1,567 @@
+from inspect import isfunction
+import math
+import torch
+import torch.nn.functional as F
+from torch import nn, einsum
+from einops import rearrange, repeat
+from typing import Optional, Any
+
+from .diffusionmodules.util import checkpoint
+from .sub_quadratic_attention import efficient_dot_product_attention
+
+from fcbh import model_management
+
+if model_management.xformers_enabled():
+    import xformers
+    import xformers.ops
+
+from fcbh.cli_args import args
+import fcbh.ops
+
+# CrossAttn precision handling
+if args.dont_upcast_attention:
+    print("disabling upcasting of attention")
+    _ATTN_PRECISION = "fp16"
+else:
+    _ATTN_PRECISION = "fp32"
+
+
+def exists(val):
+    return val is not None
+
+
+def uniq(arr):
+    return{el: True for el in arr}.keys()
+
+
+def default(val, d):
+    if exists(val):
+        return val
+    return d
+
+
+def max_neg_value(t):
+    return -torch.finfo(t.dtype).max
+
+
+def init_(tensor):
+    dim = tensor.shape[-1]
+    std = 1 / math.sqrt(dim)
+    tensor.uniform_(-std, std)
+    return tensor
+
+
+# feedforward
+class GEGLU(nn.Module):
+    def __init__(self, dim_in, dim_out, dtype=None, device=None, operations=fcbh.ops):
+        super().__init__()
+        self.proj = operations.Linear(dim_in, dim_out * 2, dtype=dtype, device=device)
+
+    def forward(self, x):
+        x, gate = self.proj(x).chunk(2, dim=-1)
+        return x * F.gelu(gate)
+
+
+class FeedForward(nn.Module):
+    def __init__(self, dim, dim_out=None, mult=4, glu=False, dropout=0., dtype=None, device=None, operations=fcbh.ops):
+        super().__init__()
+        inner_dim = int(dim * mult)
+        dim_out = default(dim_out, dim)
+        project_in = nn.Sequential(
+            operations.Linear(dim, inner_dim, dtype=dtype, device=device),
+            nn.GELU()
+        ) if not glu else GEGLU(dim, inner_dim, dtype=dtype, device=device, operations=operations)
+
+        self.net = nn.Sequential(
+            project_in,
+            nn.Dropout(dropout),
+            operations.Linear(inner_dim, dim_out, dtype=dtype, device=device)
+        )
+
+    def forward(self, x):
+        return self.net(x)
+
+
+def zero_module(module):
+    """
+    Zero out the parameters of a module and return it.
+    """
+    for p in module.parameters():
+        p.detach().zero_()
+    return module
+
+
+def Normalize(in_channels, dtype=None, device=None):
+    return torch.nn.GroupNorm(num_groups=32, num_channels=in_channels, eps=1e-6, affine=True, dtype=dtype, device=device)
+
+def attention_basic(q, k, v, heads, mask=None):
+    b, _, dim_head = q.shape
+    dim_head //= heads
+    scale = dim_head ** -0.5
+
+    h = heads
+    q, k, v = map(
+        lambda t: t.unsqueeze(3)
+        .reshape(b, -1, heads, dim_head)
+        .permute(0, 2, 1, 3)
+        .reshape(b * heads, -1, dim_head)
+        .contiguous(),
+        (q, k, v),
+    )
+
+    # force cast to fp32 to avoid overflowing
+    if _ATTN_PRECISION =="fp32":
+        with torch.autocast(enabled=False, device_type = 'cuda'):
+            q, k = q.float(), k.float()
+            sim = einsum('b i d, b j d -> b i j', q, k) * scale
+    else:
+        sim = einsum('b i d, b j d -> b i j', q, k) * scale
+
+    del q, k
+
+    if exists(mask):
+        mask = rearrange(mask, 'b ... -> b (...)')
+        max_neg_value = -torch.finfo(sim.dtype).max
+        mask = repeat(mask, 'b j -> (b h) () j', h=h)
+        sim.masked_fill_(~mask, max_neg_value)
+
+    # attention, what we cannot get enough of
+    sim = sim.softmax(dim=-1)
+
+    out = einsum('b i j, b j d -> b i d', sim.to(v.dtype), v)
+    out = (
+        out.unsqueeze(0)
+        .reshape(b, heads, -1, dim_head)
+        .permute(0, 2, 1, 3)
+        .reshape(b, -1, heads * dim_head)
+    )
+    return out
+
+
+def attention_sub_quad(query, key, value, heads, mask=None):
+    b, _, dim_head = query.shape
+    dim_head //= heads
+
+    scale = dim_head ** -0.5
+    query = query.unsqueeze(3).reshape(b, -1, heads, dim_head).permute(0, 2, 1, 3).reshape(b * heads, -1, dim_head)
+    value = value.unsqueeze(3).reshape(b, -1, heads, dim_head).permute(0, 2, 1, 3).reshape(b * heads, -1, dim_head)
+
+    key = key.unsqueeze(3).reshape(b, -1, heads, dim_head).permute(0, 2, 3, 1).reshape(b * heads, dim_head, -1)
+
+    dtype = query.dtype
+    upcast_attention = _ATTN_PRECISION =="fp32" and query.dtype != torch.float32
+    if upcast_attention:
+        bytes_per_token = torch.finfo(torch.float32).bits//8
+    else:
+        bytes_per_token = torch.finfo(query.dtype).bits//8
+    batch_x_heads, q_tokens, _ = query.shape
+    _, _, k_tokens = key.shape
+    qk_matmul_size_bytes = batch_x_heads * bytes_per_token * q_tokens * k_tokens
+
+    mem_free_total, mem_free_torch = model_management.get_free_memory(query.device, True)
+
+    kv_chunk_size_min = None
+    kv_chunk_size = None
+    query_chunk_size = None
+
+    for x in [4096, 2048, 1024, 512, 256]:
+        count = mem_free_total / (batch_x_heads * bytes_per_token * x * 4.0)
+        if count >= k_tokens:
+            kv_chunk_size = k_tokens
+            query_chunk_size = x
+            break
+
+    if query_chunk_size is None:
+        query_chunk_size = 512
+
+    hidden_states = efficient_dot_product_attention(
+        query,
+        key,
+        value,
+        query_chunk_size=query_chunk_size,
+        kv_chunk_size=kv_chunk_size,
+        kv_chunk_size_min=kv_chunk_size_min,
+        use_checkpoint=False,
+        upcast_attention=upcast_attention,
+    )
+
+    hidden_states = hidden_states.to(dtype)
+
+    hidden_states = hidden_states.unflatten(0, (-1, heads)).transpose(1,2).flatten(start_dim=2)
+    return hidden_states
+
+def attention_split(q, k, v, heads, mask=None):
+    b, _, dim_head = q.shape
+    dim_head //= heads
+    scale = dim_head ** -0.5
+
+    h = heads
+    q, k, v = map(
+        lambda t: t.unsqueeze(3)
+        .reshape(b, -1, heads, dim_head)
+        .permute(0, 2, 1, 3)
+        .reshape(b * heads, -1, dim_head)
+        .contiguous(),
+        (q, k, v),
+    )
+
+    r1 = torch.zeros(q.shape[0], q.shape[1], v.shape[2], device=q.device, dtype=q.dtype)
+
+    mem_free_total = model_management.get_free_memory(q.device)
+
+    if _ATTN_PRECISION =="fp32":
+        element_size = 4
+    else:
+        element_size = q.element_size()
+
+    gb = 1024 ** 3
+    tensor_size = q.shape[0] * q.shape[1] * k.shape[1] * element_size
+    modifier = 3
+    mem_required = tensor_size * modifier
+    steps = 1
+
+
+    if mem_required > mem_free_total:
+        steps = 2**(math.ceil(math.log(mem_required / mem_free_total, 2)))
+        # print(f"Expected tensor size:{tensor_size/gb:0.1f}GB, cuda free:{mem_free_cuda/gb:0.1f}GB "
+        #      f"torch free:{mem_free_torch/gb:0.1f} total:{mem_free_total/gb:0.1f} steps:{steps}")
+
+    if steps > 64:
+        max_res = math.floor(math.sqrt(math.sqrt(mem_free_total / 2.5)) / 8) * 64
+        raise RuntimeError(f'Not enough memory, use lower resolution (max approx. {max_res}x{max_res}). '
+                            f'Need: {mem_required/64/gb:0.1f}GB free, Have:{mem_free_total/gb:0.1f}GB free')
+
+    # print("steps", steps, mem_required, mem_free_total, modifier, q.element_size(), tensor_size)
+    first_op_done = False
+    cleared_cache = False
+    while True:
+        try:
+            slice_size = q.shape[1] // steps if (q.shape[1] % steps) == 0 else q.shape[1]
+            for i in range(0, q.shape[1], slice_size):
+                end = i + slice_size
+                if _ATTN_PRECISION =="fp32":
+                    with torch.autocast(enabled=False, device_type = 'cuda'):
+                        s1 = einsum('b i d, b j d -> b i j', q[:, i:end].float(), k.float()) * scale
+                else:
+                    s1 = einsum('b i d, b j d -> b i j', q[:, i:end], k) * scale
+
+                s2 = s1.softmax(dim=-1).to(v.dtype)
+                del s1
+                first_op_done = True
+
+                r1[:, i:end] = einsum('b i j, b j d -> b i d', s2, v)
+                del s2
+            break
+        except model_management.OOM_EXCEPTION as e:
+            if first_op_done == False:
+                model_management.soft_empty_cache(True)
+                if cleared_cache == False:
+                    cleared_cache = True
+                    print("out of memory error, emptying cache and trying again")
+                    continue
+                steps *= 2
+                if steps > 64:
+                    raise e
+                print("out of memory error, increasing steps and trying again", steps)
+            else:
+                raise e
+
+    del q, k, v
+
+    r1 = (
+        r1.unsqueeze(0)
+        .reshape(b, heads, -1, dim_head)
+        .permute(0, 2, 1, 3)
+        .reshape(b, -1, heads * dim_head)
+    )
+    return r1
+
+def attention_xformers(q, k, v, heads, mask=None):
+    b, _, dim_head = q.shape
+    dim_head //= heads
+
+    q, k, v = map(
+        lambda t: t.unsqueeze(3)
+        .reshape(b, -1, heads, dim_head)
+        .permute(0, 2, 1, 3)
+        .reshape(b * heads, -1, dim_head)
+        .contiguous(),
+        (q, k, v),
+    )
+
+    # actually compute the attention, what we cannot get enough of
+    out = xformers.ops.memory_efficient_attention(q, k, v, attn_bias=None)
+
+    if exists(mask):
+        raise NotImplementedError
+    out = (
+        out.unsqueeze(0)
+        .reshape(b, heads, -1, dim_head)
+        .permute(0, 2, 1, 3)
+        .reshape(b, -1, heads * dim_head)
+    )
+    return out
+
+def attention_pytorch(q, k, v, heads, mask=None):
+    b, _, dim_head = q.shape
+    dim_head //= heads
+    q, k, v = map(
+        lambda t: t.view(b, -1, heads, dim_head).transpose(1, 2),
+        (q, k, v),
+    )
+
+    out = torch.nn.functional.scaled_dot_product_attention(q, k, v, attn_mask=mask, dropout_p=0.0, is_causal=False)
+    out = (
+        out.transpose(1, 2).reshape(b, -1, heads * dim_head)
+    )
+    return out
+
+
+optimized_attention = attention_basic
+optimized_attention_masked = attention_basic
+
+if model_management.xformers_enabled():
+    print("Using xformers cross attention")
+    optimized_attention = attention_xformers
+elif model_management.pytorch_attention_enabled():
+    print("Using pytorch cross attention")
+    optimized_attention = attention_pytorch
+else:
+    if args.use_split_cross_attention:
+        print("Using split optimization for cross attention")
+        optimized_attention = attention_split
+    else:
+        print("Using sub quadratic optimization for cross attention, if you have memory or speed issues try using: --use-split-cross-attention")
+        optimized_attention = attention_sub_quad
+
+if model_management.pytorch_attention_enabled():
+    optimized_attention_masked = attention_pytorch
+
+class CrossAttention(nn.Module):
+    def __init__(self, query_dim, context_dim=None, heads=8, dim_head=64, dropout=0., dtype=None, device=None, operations=fcbh.ops):
+        super().__init__()
+        inner_dim = dim_head * heads
+        context_dim = default(context_dim, query_dim)
+
+        self.heads = heads
+        self.dim_head = dim_head
+
+        self.to_q = operations.Linear(query_dim, inner_dim, bias=False, dtype=dtype, device=device)
+        self.to_k = operations.Linear(context_dim, inner_dim, bias=False, dtype=dtype, device=device)
+        self.to_v = operations.Linear(context_dim, inner_dim, bias=False, dtype=dtype, device=device)
+
+        self.to_out = nn.Sequential(operations.Linear(inner_dim, query_dim, dtype=dtype, device=device), nn.Dropout(dropout))
+
+    def forward(self, x, context=None, value=None, mask=None):
+        q = self.to_q(x)
+        context = default(context, x)
+        k = self.to_k(context)
+        if value is not None:
+            v = self.to_v(value)
+            del value
+        else:
+            v = self.to_v(context)
+
+        if mask is None:
+            out = optimized_attention(q, k, v, self.heads)
+        else:
+            out = optimized_attention_masked(q, k, v, self.heads, mask)
+        return self.to_out(out)
+
+
+class BasicTransformerBlock(nn.Module):
+    def __init__(self, dim, n_heads, d_head, dropout=0., context_dim=None, gated_ff=True, checkpoint=True,
+                 disable_self_attn=False, dtype=None, device=None, operations=fcbh.ops):
+        super().__init__()
+        self.disable_self_attn = disable_self_attn
+        self.attn1 = CrossAttention(query_dim=dim, heads=n_heads, dim_head=d_head, dropout=dropout,
+                              context_dim=context_dim if self.disable_self_attn else None, dtype=dtype, device=device, operations=operations)  # is a self-attention if not self.disable_self_attn
+        self.ff = FeedForward(dim, dropout=dropout, glu=gated_ff, dtype=dtype, device=device, operations=operations)
+        self.attn2 = CrossAttention(query_dim=dim, context_dim=context_dim,
+                              heads=n_heads, dim_head=d_head, dropout=dropout, dtype=dtype, device=device, operations=operations)  # is self-attn if context is none
+        self.norm1 = nn.LayerNorm(dim, dtype=dtype, device=device)
+        self.norm2 = nn.LayerNorm(dim, dtype=dtype, device=device)
+        self.norm3 = nn.LayerNorm(dim, dtype=dtype, device=device)
+        self.checkpoint = checkpoint
+        self.n_heads = n_heads
+        self.d_head = d_head
+
+    def forward(self, x, context=None, transformer_options={}):
+        return checkpoint(self._forward, (x, context, transformer_options), self.parameters(), self.checkpoint)
+
+    def _forward(self, x, context=None, transformer_options={}):
+        extra_options = {}
+        block = None
+        block_index = 0
+        if "current_index" in transformer_options:
+            extra_options["transformer_index"] = transformer_options["current_index"]
+        if "block_index" in transformer_options:
+            block_index = transformer_options["block_index"]
+            extra_options["block_index"] = block_index
+        if "original_shape" in transformer_options:
+            extra_options["original_shape"] = transformer_options["original_shape"]
+        if "block" in transformer_options:
+            block = transformer_options["block"]
+            extra_options["block"] = block
+        if "cond_or_uncond" in transformer_options:
+            extra_options["cond_or_uncond"] = transformer_options["cond_or_uncond"]
+        if "patches" in transformer_options:
+            transformer_patches = transformer_options["patches"]
+        else:
+            transformer_patches = {}
+
+        extra_options["n_heads"] = self.n_heads
+        extra_options["dim_head"] = self.d_head
+
+        if "patches_replace" in transformer_options:
+            transformer_patches_replace = transformer_options["patches_replace"]
+        else:
+            transformer_patches_replace = {}
+
+        n = self.norm1(x)
+        if self.disable_self_attn:
+            context_attn1 = context
+        else:
+            context_attn1 = None
+        value_attn1 = None
+
+        if "attn1_patch" in transformer_patches:
+            patch = transformer_patches["attn1_patch"]
+            if context_attn1 is None:
+                context_attn1 = n
+            value_attn1 = context_attn1
+            for p in patch:
+                n, context_attn1, value_attn1 = p(n, context_attn1, value_attn1, extra_options)
+
+        if block is not None:
+            transformer_block = (block[0], block[1], block_index)
+        else:
+            transformer_block = None
+        attn1_replace_patch = transformer_patches_replace.get("attn1", {})
+        block_attn1 = transformer_block
+        if block_attn1 not in attn1_replace_patch:
+            block_attn1 = block
+
+        if block_attn1 in attn1_replace_patch:
+            if context_attn1 is None:
+                context_attn1 = n
+                value_attn1 = n
+            n = self.attn1.to_q(n)
+            context_attn1 = self.attn1.to_k(context_attn1)
+            value_attn1 = self.attn1.to_v(value_attn1)
+            n = attn1_replace_patch[block_attn1](n, context_attn1, value_attn1, extra_options)
+            n = self.attn1.to_out(n)
+        else:
+            n = self.attn1(n, context=context_attn1, value=value_attn1)
+
+        if "attn1_output_patch" in transformer_patches:
+            patch = transformer_patches["attn1_output_patch"]
+            for p in patch:
+                n = p(n, extra_options)
+
+        x += n
+        if "middle_patch" in transformer_patches:
+            patch = transformer_patches["middle_patch"]
+            for p in patch:
+                x = p(x, extra_options)
+
+        n = self.norm2(x)
+
+        context_attn2 = context
+        value_attn2 = None
+        if "attn2_patch" in transformer_patches:
+            patch = transformer_patches["attn2_patch"]
+            value_attn2 = context_attn2
+            for p in patch:
+                n, context_attn2, value_attn2 = p(n, context_attn2, value_attn2, extra_options)
+
+        attn2_replace_patch = transformer_patches_replace.get("attn2", {})
+        block_attn2 = transformer_block
+        if block_attn2 not in attn2_replace_patch:
+            block_attn2 = block
+
+        if block_attn2 in attn2_replace_patch:
+            if value_attn2 is None:
+                value_attn2 = context_attn2
+            n = self.attn2.to_q(n)
+            context_attn2 = self.attn2.to_k(context_attn2)
+            value_attn2 = self.attn2.to_v(value_attn2)
+            n = attn2_replace_patch[block_attn2](n, context_attn2, value_attn2, extra_options)
+            n = self.attn2.to_out(n)
+        else:
+            n = self.attn2(n, context=context_attn2, value=value_attn2)
+
+        if "attn2_output_patch" in transformer_patches:
+            patch = transformer_patches["attn2_output_patch"]
+            for p in patch:
+                n = p(n, extra_options)
+
+        x += n
+        x = self.ff(self.norm3(x)) + x
+        return x
+
+
+class SpatialTransformer(nn.Module):
+    """
+    Transformer block for image-like data.
+    First, project the input (aka embedding)
+    and reshape to b, t, d.
+    Then apply standard transformer action.
+    Finally, reshape to image
+    NEW: use_linear for more efficiency instead of the 1x1 convs
+    """
+    def __init__(self, in_channels, n_heads, d_head,
+                 depth=1, dropout=0., context_dim=None,
+                 disable_self_attn=False, use_linear=False,
+                 use_checkpoint=True, dtype=None, device=None, operations=fcbh.ops):
+        super().__init__()
+        if exists(context_dim) and not isinstance(context_dim, list):
+            context_dim = [context_dim] * depth
+        self.in_channels = in_channels
+        inner_dim = n_heads * d_head
+        self.norm = Normalize(in_channels, dtype=dtype, device=device)
+        if not use_linear:
+            self.proj_in = operations.Conv2d(in_channels,
+                                     inner_dim,
+                                     kernel_size=1,
+                                     stride=1,
+                                     padding=0, dtype=dtype, device=device)
+        else:
+            self.proj_in = operations.Linear(in_channels, inner_dim, dtype=dtype, device=device)
+
+        self.transformer_blocks = nn.ModuleList(
+            [BasicTransformerBlock(inner_dim, n_heads, d_head, dropout=dropout, context_dim=context_dim[d],
+                                   disable_self_attn=disable_self_attn, checkpoint=use_checkpoint, dtype=dtype, device=device, operations=operations)
+                for d in range(depth)]
+        )
+        if not use_linear:
+            self.proj_out = operations.Conv2d(inner_dim,in_channels,
+                                                  kernel_size=1,
+                                                  stride=1,
+                                                  padding=0, dtype=dtype, device=device)
+        else:
+            self.proj_out = operations.Linear(in_channels, inner_dim, dtype=dtype, device=device)
+        self.use_linear = use_linear
+
+    def forward(self, x, context=None, transformer_options={}):
+        # note: if no context is given, cross-attention defaults to self-attention
+        if not isinstance(context, list):
+            context = [context] * len(self.transformer_blocks)
+        b, c, h, w = x.shape
+        x_in = x
+        x = self.norm(x)
+        if not self.use_linear:
+            x = self.proj_in(x)
+        x = rearrange(x, 'b c h w -> b (h w) c').contiguous()
+        if self.use_linear:
+            x = self.proj_in(x)
+        for i, block in enumerate(self.transformer_blocks):
+            transformer_options["block_index"] = i
+            x = block(x, context=context[i], transformer_options=transformer_options)
+        if self.use_linear:
+            x = self.proj_out(x)
+        x = rearrange(x, 'b (h w) c -> b c h w', h=h, w=w).contiguous()
+        if not self.use_linear:
+            x = self.proj_out(x)
+        return x + x_in
+

backend/headless/fcbh/ldm/modules/diffusionmodules/model.py

@@ -0,0 +1,649 @@
+# pytorch_diffusion + derived encoder decoder
+import math
+import torch
+import torch.nn as nn
+import numpy as np
+from einops import rearrange
+from typing import Optional, Any
+
+from fcbh import model_management
+import fcbh.ops
+
+if model_management.xformers_enabled_vae():
+    import xformers
+    import xformers.ops
+
+def get_timestep_embedding(timesteps, embedding_dim):
+    """
+    This matches the implementation in Denoising Diffusion Probabilistic Models:
+    From Fairseq.
+    Build sinusoidal embeddings.
+    This matches the implementation in tensor2tensor, but differs slightly
+    from the description in Section 3.5 of "Attention Is All You Need".
+    """
+    assert len(timesteps.shape) == 1
+
+    half_dim = embedding_dim // 2
+    emb = math.log(10000) / (half_dim - 1)
+    emb = torch.exp(torch.arange(half_dim, dtype=torch.float32) * -emb)
+    emb = emb.to(device=timesteps.device)
+    emb = timesteps.float()[:, None] * emb[None, :]
+    emb = torch.cat([torch.sin(emb), torch.cos(emb)], dim=1)
+    if embedding_dim % 2 == 1:  # zero pad
+        emb = torch.nn.functional.pad(emb, (0,1,0,0))
+    return emb
+
+
+def nonlinearity(x):
+    # swish
+    return x*torch.sigmoid(x)
+
+
+def Normalize(in_channels, num_groups=32):
+    return torch.nn.GroupNorm(num_groups=num_groups, num_channels=in_channels, eps=1e-6, affine=True)
+
+
+class Upsample(nn.Module):
+    def __init__(self, in_channels, with_conv):
+        super().__init__()
+        self.with_conv = with_conv
+        if self.with_conv:
+            self.conv = fcbh.ops.Conv2d(in_channels,
+                                        in_channels,
+                                        kernel_size=3,
+                                        stride=1,
+                                        padding=1)
+
+    def forward(self, x):
+        try:
+            x = torch.nn.functional.interpolate(x, scale_factor=2.0, mode="nearest")
+        except: #operation not implemented for bf16
+            b, c, h, w = x.shape
+            out = torch.empty((b, c, h*2, w*2), dtype=x.dtype, layout=x.layout, device=x.device)
+            split = 8
+            l = out.shape[1] // split
+            for i in range(0, out.shape[1], l):
+                out[:,i:i+l] = torch.nn.functional.interpolate(x[:,i:i+l].to(torch.float32), scale_factor=2.0, mode="nearest").to(x.dtype)
+            del x
+            x = out
+
+        if self.with_conv:
+            x = self.conv(x)
+        return x
+
+
+class Downsample(nn.Module):
+    def __init__(self, in_channels, with_conv):
+        super().__init__()
+        self.with_conv = with_conv
+        if self.with_conv:
+            # no asymmetric padding in torch conv, must do it ourselves
+            self.conv = fcbh.ops.Conv2d(in_channels,
+                                        in_channels,
+                                        kernel_size=3,
+                                        stride=2,
+                                        padding=0)
+
+    def forward(self, x):
+        if self.with_conv:
+            pad = (0,1,0,1)
+            x = torch.nn.functional.pad(x, pad, mode="constant", value=0)
+            x = self.conv(x)
+        else:
+            x = torch.nn.functional.avg_pool2d(x, kernel_size=2, stride=2)
+        return x
+
+
+class ResnetBlock(nn.Module):
+    def __init__(self, *, in_channels, out_channels=None, conv_shortcut=False,
+                 dropout, temb_channels=512):
+        super().__init__()
+        self.in_channels = in_channels
+        out_channels = in_channels if out_channels is None else out_channels
+        self.out_channels = out_channels
+        self.use_conv_shortcut = conv_shortcut
+
+        self.swish = torch.nn.SiLU(inplace=True)
+        self.norm1 = Normalize(in_channels)
+        self.conv1 = fcbh.ops.Conv2d(in_channels,
+                                     out_channels,
+                                     kernel_size=3,
+                                     stride=1,
+                                     padding=1)
+        if temb_channels > 0:
+            self.temb_proj = fcbh.ops.Linear(temb_channels,
+                                             out_channels)
+        self.norm2 = Normalize(out_channels)
+        self.dropout = torch.nn.Dropout(dropout, inplace=True)
+        self.conv2 = fcbh.ops.Conv2d(out_channels,
+                                     out_channels,
+                                     kernel_size=3,
+                                     stride=1,
+                                     padding=1)
+        if self.in_channels != self.out_channels:
+            if self.use_conv_shortcut:
+                self.conv_shortcut = fcbh.ops.Conv2d(in_channels,
+                                                     out_channels,
+                                                     kernel_size=3,
+                                                     stride=1,
+                                                     padding=1)
+            else:
+                self.nin_shortcut = fcbh.ops.Conv2d(in_channels,
+                                                    out_channels,
+                                                    kernel_size=1,
+                                                    stride=1,
+                                                    padding=0)
+
+    def forward(self, x, temb):
+        h = x
+        h = self.norm1(h)
+        h = self.swish(h)
+        h = self.conv1(h)
+
+        if temb is not None:
+            h = h + self.temb_proj(self.swish(temb))[:,:,None,None]
+
+        h = self.norm2(h)
+        h = self.swish(h)
+        h = self.dropout(h)
+        h = self.conv2(h)
+
+        if self.in_channels != self.out_channels:
+            if self.use_conv_shortcut:
+                x = self.conv_shortcut(x)
+            else:
+                x = self.nin_shortcut(x)
+
+        return x+h
+
+def slice_attention(q, k, v):
+    r1 = torch.zeros_like(k, device=q.device)
+    scale = (int(q.shape[-1])**(-0.5))
+
+    mem_free_total = model_management.get_free_memory(q.device)
+
+    gb = 1024 ** 3
+    tensor_size = q.shape[0] * q.shape[1] * k.shape[2] * q.element_size()
+    modifier = 3 if q.element_size() == 2 else 2.5
+    mem_required = tensor_size * modifier
+    steps = 1
+
+    if mem_required > mem_free_total:
+        steps = 2**(math.ceil(math.log(mem_required / mem_free_total, 2)))
+
+    while True:
+        try:
+            slice_size = q.shape[1] // steps if (q.shape[1] % steps) == 0 else q.shape[1]
+            for i in range(0, q.shape[1], slice_size):
+                end = i + slice_size
+                s1 = torch.bmm(q[:, i:end], k) * scale
+
+                s2 = torch.nn.functional.softmax(s1, dim=2).permute(0,2,1)
+                del s1
+
+                r1[:, :, i:end] = torch.bmm(v, s2)
+                del s2
+            break
+        except model_management.OOM_EXCEPTION as e:
+            model_management.soft_empty_cache(True)
+            steps *= 2
+            if steps > 128:
+                raise e
+            print("out of memory error, increasing steps and trying again", steps)
+
+    return r1
+
+def normal_attention(q, k, v):
+    # compute attention
+    b,c,h,w = q.shape
+
+    q = q.reshape(b,c,h*w)
+    q = q.permute(0,2,1)   # b,hw,c
+    k = k.reshape(b,c,h*w) # b,c,hw
+    v = v.reshape(b,c,h*w)
+
+    r1 = slice_attention(q, k, v)
+    h_ = r1.reshape(b,c,h,w)
+    del r1
+    return h_
+
+def xformers_attention(q, k, v):
+    # compute attention
+    B, C, H, W = q.shape
+    q, k, v = map(
+        lambda t: t.view(B, C, -1).transpose(1, 2).contiguous(),
+        (q, k, v),
+    )
+
+    try:
+        out = xformers.ops.memory_efficient_attention(q, k, v, attn_bias=None)
+        out = out.transpose(1, 2).reshape(B, C, H, W)
+    except NotImplementedError as e:
+        out = slice_attention(q.view(B, -1, C), k.view(B, -1, C).transpose(1, 2), v.view(B, -1, C).transpose(1, 2)).reshape(B, C, H, W)
+    return out
+
+def pytorch_attention(q, k, v):
+    # compute attention
+    B, C, H, W = q.shape
+    q, k, v = map(
+        lambda t: t.view(B, 1, C, -1).transpose(2, 3).contiguous(),
+        (q, k, v),
+    )
+
+    try:
+        out = torch.nn.functional.scaled_dot_product_attention(q, k, v, attn_mask=None, dropout_p=0.0, is_causal=False)
+        out = out.transpose(2, 3).reshape(B, C, H, W)
+    except model_management.OOM_EXCEPTION as e:
+        print("scaled_dot_product_attention OOMed: switched to slice attention")
+        out = slice_attention(q.view(B, -1, C), k.view(B, -1, C).transpose(1, 2), v.view(B, -1, C).transpose(1, 2)).reshape(B, C, H, W)
+    return out
+
+
+class AttnBlock(nn.Module):
+    def __init__(self, in_channels):
+        super().__init__()
+        self.in_channels = in_channels
+
+        self.norm = Normalize(in_channels)
+        self.q = fcbh.ops.Conv2d(in_channels,
+                                 in_channels,
+                                 kernel_size=1,
+                                 stride=1,
+                                 padding=0)
+        self.k = fcbh.ops.Conv2d(in_channels,
+                                 in_channels,
+                                 kernel_size=1,
+                                 stride=1,
+                                 padding=0)
+        self.v = fcbh.ops.Conv2d(in_channels,
+                                 in_channels,
+                                 kernel_size=1,
+                                 stride=1,
+                                 padding=0)
+        self.proj_out = fcbh.ops.Conv2d(in_channels,
+                                        in_channels,
+                                        kernel_size=1,
+                                        stride=1,
+                                        padding=0)
+
+        if model_management.xformers_enabled_vae():
+            print("Using xformers attention in VAE")
+            self.optimized_attention = xformers_attention
+        elif model_management.pytorch_attention_enabled():
+            print("Using pytorch attention in VAE")
+            self.optimized_attention = pytorch_attention
+        else:
+            print("Using split attention in VAE")
+            self.optimized_attention = normal_attention
+
+    def forward(self, x):
+        h_ = x
+        h_ = self.norm(h_)
+        q = self.q(h_)
+        k = self.k(h_)
+        v = self.v(h_)
+
+        h_ = self.optimized_attention(q, k, v)
+
+        h_ = self.proj_out(h_)
+
+        return x+h_
+
+
+def make_attn(in_channels, attn_type="vanilla", attn_kwargs=None):
+    return AttnBlock(in_channels)
+
+
+class Model(nn.Module):
+    def __init__(self, *, ch, out_ch, ch_mult=(1,2,4,8), num_res_blocks,
+                 attn_resolutions, dropout=0.0, resamp_with_conv=True, in_channels,
+                 resolution, use_timestep=True, use_linear_attn=False, attn_type="vanilla"):
+        super().__init__()
+        if use_linear_attn: attn_type = "linear"
+        self.ch = ch
+        self.temb_ch = self.ch*4
+        self.num_resolutions = len(ch_mult)
+        self.num_res_blocks = num_res_blocks
+        self.resolution = resolution
+        self.in_channels = in_channels
+
+        self.use_timestep = use_timestep
+        if self.use_timestep:
+            # timestep embedding
+            self.temb = nn.Module()
+            self.temb.dense = nn.ModuleList([
+                fcbh.ops.Linear(self.ch,
+                                self.temb_ch),
+                fcbh.ops.Linear(self.temb_ch,
+                                self.temb_ch),
+            ])
+
+        # downsampling
+        self.conv_in = fcbh.ops.Conv2d(in_channels,
+                                       self.ch,
+                                       kernel_size=3,
+                                       stride=1,
+                                       padding=1)
+
+        curr_res = resolution
+        in_ch_mult = (1,)+tuple(ch_mult)
+        self.down = nn.ModuleList()
+        for i_level in range(self.num_resolutions):
+            block = nn.ModuleList()
+            attn = nn.ModuleList()
+            block_in = ch*in_ch_mult[i_level]
+            block_out = ch*ch_mult[i_level]
+            for i_block in range(self.num_res_blocks):
+                block.append(ResnetBlock(in_channels=block_in,
+                                         out_channels=block_out,
+                                         temb_channels=self.temb_ch,
+                                         dropout=dropout))
+                block_in = block_out
+                if curr_res in attn_resolutions:
+                    attn.append(make_attn(block_in, attn_type=attn_type))
+            down = nn.Module()
+            down.block = block
+            down.attn = attn
+            if i_level != self.num_resolutions-1:
+                down.downsample = Downsample(block_in, resamp_with_conv)
+                curr_res = curr_res // 2
+            self.down.append(down)
+
+        # middle
+        self.mid = nn.Module()
+        self.mid.block_1 = ResnetBlock(in_channels=block_in,
+                                       out_channels=block_in,
+                                       temb_channels=self.temb_ch,
+                                       dropout=dropout)
+        self.mid.attn_1 = make_attn(block_in, attn_type=attn_type)
+        self.mid.block_2 = ResnetBlock(in_channels=block_in,
+                                       out_channels=block_in,
+                                       temb_channels=self.temb_ch,
+                                       dropout=dropout)
+
+        # upsampling
+        self.up = nn.ModuleList()
+        for i_level in reversed(range(self.num_resolutions)):
+            block = nn.ModuleList()
+            attn = nn.ModuleList()
+            block_out = ch*ch_mult[i_level]
+            skip_in = ch*ch_mult[i_level]
+            for i_block in range(self.num_res_blocks+1):
+                if i_block == self.num_res_blocks:
+                    skip_in = ch*in_ch_mult[i_level]
+                block.append(ResnetBlock(in_channels=block_in+skip_in,
+                                         out_channels=block_out,
+                                         temb_channels=self.temb_ch,
+                                         dropout=dropout))
+                block_in = block_out
+                if curr_res in attn_resolutions:
+                    attn.append(make_attn(block_in, attn_type=attn_type))
+            up = nn.Module()
+            up.block = block
+            up.attn = attn
+            if i_level != 0:
+                up.upsample = Upsample(block_in, resamp_with_conv)
+                curr_res = curr_res * 2
+            self.up.insert(0, up) # prepend to get consistent order
+
+        # end
+        self.norm_out = Normalize(block_in)
+        self.conv_out = fcbh.ops.Conv2d(block_in,
+                                        out_ch,
+                                        kernel_size=3,
+                                        stride=1,
+                                        padding=1)
+
+    def forward(self, x, t=None, context=None):
+        #assert x.shape[2] == x.shape[3] == self.resolution
+        if context is not None:
+            # assume aligned context, cat along channel axis
+            x = torch.cat((x, context), dim=1)
+        if self.use_timestep:
+            # timestep embedding
+            assert t is not None
+            temb = get_timestep_embedding(t, self.ch)
+            temb = self.temb.dense[0](temb)
+            temb = nonlinearity(temb)
+            temb = self.temb.dense[1](temb)
+        else:
+            temb = None
+
+        # downsampling
+        hs = [self.conv_in(x)]
+        for i_level in range(self.num_resolutions):
+            for i_block in range(self.num_res_blocks):
+                h = self.down[i_level].block[i_block](hs[-1], temb)
+                if len(self.down[i_level].attn) > 0:
+                    h = self.down[i_level].attn[i_block](h)
+                hs.append(h)
+            if i_level != self.num_resolutions-1:
+                hs.append(self.down[i_level].downsample(hs[-1]))
+
+        # middle
+        h = hs[-1]
+        h = self.mid.block_1(h, temb)
+        h = self.mid.attn_1(h)
+        h = self.mid.block_2(h, temb)
+
+        # upsampling
+        for i_level in reversed(range(self.num_resolutions)):
+            for i_block in range(self.num_res_blocks+1):
+                h = self.up[i_level].block[i_block](
+                    torch.cat([h, hs.pop()], dim=1), temb)
+                if len(self.up[i_level].attn) > 0:
+                    h = self.up[i_level].attn[i_block](h)
+            if i_level != 0:
+                h = self.up[i_level].upsample(h)
+
+        # end
+        h = self.norm_out(h)
+        h = nonlinearity(h)
+        h = self.conv_out(h)
+        return h
+
+    def get_last_layer(self):
+        return self.conv_out.weight
+
+
+class Encoder(nn.Module):
+    def __init__(self, *, ch, out_ch, ch_mult=(1,2,4,8), num_res_blocks,
+                 attn_resolutions, dropout=0.0, resamp_with_conv=True, in_channels,
+                 resolution, z_channels, double_z=True, use_linear_attn=False, attn_type="vanilla",
+                 **ignore_kwargs):
+        super().__init__()
+        if use_linear_attn: attn_type = "linear"
+        self.ch = ch
+        self.temb_ch = 0
+        self.num_resolutions = len(ch_mult)
+        self.num_res_blocks = num_res_blocks
+        self.resolution = resolution
+        self.in_channels = in_channels
+
+        # downsampling
+        self.conv_in = fcbh.ops.Conv2d(in_channels,
+                                       self.ch,
+                                       kernel_size=3,
+                                       stride=1,
+                                       padding=1)
+
+        curr_res = resolution
+        in_ch_mult = (1,)+tuple(ch_mult)
+        self.in_ch_mult = in_ch_mult
+        self.down = nn.ModuleList()
+        for i_level in range(self.num_resolutions):
+            block = nn.ModuleList()
+            attn = nn.ModuleList()
+            block_in = ch*in_ch_mult[i_level]
+            block_out = ch*ch_mult[i_level]
+            for i_block in range(self.num_res_blocks):
+                block.append(ResnetBlock(in_channels=block_in,
+                                         out_channels=block_out,
+                                         temb_channels=self.temb_ch,
+                                         dropout=dropout))
+                block_in = block_out
+                if curr_res in attn_resolutions:
+                    attn.append(make_attn(block_in, attn_type=attn_type))
+            down = nn.Module()
+            down.block = block
+            down.attn = attn
+            if i_level != self.num_resolutions-1:
+                down.downsample = Downsample(block_in, resamp_with_conv)
+                curr_res = curr_res // 2
+            self.down.append(down)
+
+        # middle
+        self.mid = nn.Module()
+        self.mid.block_1 = ResnetBlock(in_channels=block_in,
+                                       out_channels=block_in,
+                                       temb_channels=self.temb_ch,
+                                       dropout=dropout)
+        self.mid.attn_1 = make_attn(block_in, attn_type=attn_type)
+        self.mid.block_2 = ResnetBlock(in_channels=block_in,
+                                       out_channels=block_in,
+                                       temb_channels=self.temb_ch,
+                                       dropout=dropout)
+
+        # end
+        self.norm_out = Normalize(block_in)
+        self.conv_out = fcbh.ops.Conv2d(block_in,
+                                        2*z_channels if double_z else z_channels,
+                                        kernel_size=3,
+                                        stride=1,
+                                        padding=1)
+
+    def forward(self, x):
+        # timestep embedding
+        temb = None
+        # downsampling
+        h = self.conv_in(x)
+        for i_level in range(self.num_resolutions):
+            for i_block in range(self.num_res_blocks):
+                h = self.down[i_level].block[i_block](h, temb)
+                if len(self.down[i_level].attn) > 0:
+                    h = self.down[i_level].attn[i_block](h)
+            if i_level != self.num_resolutions-1:
+                h = self.down[i_level].downsample(h)
+
+        # middle
+        h = self.mid.block_1(h, temb)
+        h = self.mid.attn_1(h)
+        h = self.mid.block_2(h, temb)
+
+        # end
+        h = self.norm_out(h)
+        h = nonlinearity(h)
+        h = self.conv_out(h)
+        return h
+
+
+class Decoder(nn.Module):
+    def __init__(self, *, ch, out_ch, ch_mult=(1,2,4,8), num_res_blocks,
+                 attn_resolutions, dropout=0.0, resamp_with_conv=True, in_channels,
+                 resolution, z_channels, give_pre_end=False, tanh_out=False, use_linear_attn=False,
+                 conv_out_op=fcbh.ops.Conv2d,
+                 resnet_op=ResnetBlock,
+                 attn_op=AttnBlock,
+                **ignorekwargs):
+        super().__init__()
+        if use_linear_attn: attn_type = "linear"
+        self.ch = ch
+        self.temb_ch = 0
+        self.num_resolutions = len(ch_mult)
+        self.num_res_blocks = num_res_blocks
+        self.resolution = resolution
+        self.in_channels = in_channels
+        self.give_pre_end = give_pre_end
+        self.tanh_out = tanh_out
+
+        # compute in_ch_mult, block_in and curr_res at lowest res
+        in_ch_mult = (1,)+tuple(ch_mult)
+        block_in = ch*ch_mult[self.num_resolutions-1]
+        curr_res = resolution // 2**(self.num_resolutions-1)
+        self.z_shape = (1,z_channels,curr_res,curr_res)
+        print("Working with z of shape {} = {} dimensions.".format(
+            self.z_shape, np.prod(self.z_shape)))
+
+        # z to block_in
+        self.conv_in = fcbh.ops.Conv2d(z_channels,
+                                       block_in,
+                                       kernel_size=3,
+                                       stride=1,
+                                       padding=1)
+
+        # middle
+        self.mid = nn.Module()
+        self.mid.block_1 = resnet_op(in_channels=block_in,
+                                       out_channels=block_in,
+                                       temb_channels=self.temb_ch,
+                                       dropout=dropout)
+        self.mid.attn_1 = attn_op(block_in)
+        self.mid.block_2 = resnet_op(in_channels=block_in,
+                                       out_channels=block_in,
+                                       temb_channels=self.temb_ch,
+                                       dropout=dropout)
+
+        # upsampling
+        self.up = nn.ModuleList()
+        for i_level in reversed(range(self.num_resolutions)):
+            block = nn.ModuleList()
+            attn = nn.ModuleList()
+            block_out = ch*ch_mult[i_level]
+            for i_block in range(self.num_res_blocks+1):
+                block.append(resnet_op(in_channels=block_in,
+                                         out_channels=block_out,
+                                         temb_channels=self.temb_ch,
+                                         dropout=dropout))
+                block_in = block_out
+                if curr_res in attn_resolutions:
+                    attn.append(attn_op(block_in))
+            up = nn.Module()
+            up.block = block
+            up.attn = attn
+            if i_level != 0:
+                up.upsample = Upsample(block_in, resamp_with_conv)
+                curr_res = curr_res * 2
+            self.up.insert(0, up) # prepend to get consistent order
+
+        # end
+        self.norm_out = Normalize(block_in)
+        self.conv_out = conv_out_op(block_in,
+                                        out_ch,
+                                        kernel_size=3,
+                                        stride=1,
+                                        padding=1)
+
+    def forward(self, z, **kwargs):
+        #assert z.shape[1:] == self.z_shape[1:]
+        self.last_z_shape = z.shape
+
+        # timestep embedding
+        temb = None
+
+        # z to block_in
+        h = self.conv_in(z)
+
+        # middle
+        h = self.mid.block_1(h, temb, **kwargs)
+        h = self.mid.attn_1(h, **kwargs)
+        h = self.mid.block_2(h, temb, **kwargs)
+
+        # upsampling
+        for i_level in reversed(range(self.num_resolutions)):
+            for i_block in range(self.num_res_blocks+1):
+                h = self.up[i_level].block[i_block](h, temb, **kwargs)
+                if len(self.up[i_level].attn) > 0:
+                    h = self.up[i_level].attn[i_block](h, **kwargs)
+            if i_level != 0:
+                h = self.up[i_level].upsample(h)
+
+        # end
+        if self.give_pre_end:
+            return h
+
+        h = self.norm_out(h)
+        h = nonlinearity(h)
+        h = self.conv_out(h, **kwargs)
+        if self.tanh_out:
+            h = torch.tanh(h)
+        return h

backend/headless/fcbh/ldm/modules/diffusionmodules/openaimodel.py

@@ -0,0 +1,654 @@
+from abc import abstractmethod
+import math
+
+import numpy as np
+import torch as th
+import torch.nn as nn
+import torch.nn.functional as F
+
+from .util import (
+    checkpoint,
+    avg_pool_nd,
+    zero_module,
+    normalization,
+    timestep_embedding,
+)
+from ..attention import SpatialTransformer
+from fcbh.ldm.util import exists
+import fcbh.ops
+
+class TimestepBlock(nn.Module):
+    """
+    Any module where forward() takes timestep embeddings as a second argument.
+    """
+
+    @abstractmethod
+    def forward(self, x, emb):
+        """
+        Apply the module to `x` given `emb` timestep embeddings.
+        """
+
+
+class TimestepEmbedSequential(nn.Sequential, TimestepBlock):
+    """
+    A sequential module that passes timestep embeddings to the children that
+    support it as an extra input.
+    """
+
+    def forward(self, x, emb, context=None, transformer_options={}, output_shape=None):
+        for layer in self:
+            if isinstance(layer, TimestepBlock):
+                x = layer(x, emb)
+            elif isinstance(layer, SpatialTransformer):
+                x = layer(x, context, transformer_options)
+            elif isinstance(layer, Upsample):
+                x = layer(x, output_shape=output_shape)
+            else:
+                x = layer(x)
+        return x
+
+#This is needed because accelerate makes a copy of transformer_options which breaks "current_index"
+def forward_timestep_embed(ts, x, emb, context=None, transformer_options={}, output_shape=None):
+    for layer in ts:
+        if isinstance(layer, TimestepBlock):
+            x = layer(x, emb)
+        elif isinstance(layer, SpatialTransformer):
+            x = layer(x, context, transformer_options)
+            transformer_options["current_index"] += 1
+        elif isinstance(layer, Upsample):
+            x = layer(x, output_shape=output_shape)
+        else:
+            x = layer(x)
+    return x
+
+class Upsample(nn.Module):
+    """
+    An upsampling layer with an optional convolution.
+    :param channels: channels in the inputs and outputs.
+    :param use_conv: a bool determining if a convolution is applied.
+    :param dims: determines if the signal is 1D, 2D, or 3D. If 3D, then
+                 upsampling occurs in the inner-two dimensions.
+    """
+
+    def __init__(self, channels, use_conv, dims=2, out_channels=None, padding=1, dtype=None, device=None, operations=fcbh.ops):
+        super().__init__()
+        self.channels = channels
+        self.out_channels = out_channels or channels
+        self.use_conv = use_conv
+        self.dims = dims
+        if use_conv:
+            self.conv = operations.conv_nd(dims, self.channels, self.out_channels, 3, padding=padding, dtype=dtype, device=device)
+
+    def forward(self, x, output_shape=None):
+        assert x.shape[1] == self.channels
+        if self.dims == 3:
+            shape = [x.shape[2], x.shape[3] * 2, x.shape[4] * 2]
+            if output_shape is not None:
+                shape[1] = output_shape[3]
+                shape[2] = output_shape[4]
+        else:
+            shape = [x.shape[2] * 2, x.shape[3] * 2]
+            if output_shape is not None:
+                shape[0] = output_shape[2]
+                shape[1] = output_shape[3]
+
+        x = F.interpolate(x, size=shape, mode="nearest")
+        if self.use_conv:
+            x = self.conv(x)
+        return x
+
+class Downsample(nn.Module):
+    """
+    A downsampling layer with an optional convolution.
+    :param channels: channels in the inputs and outputs.
+    :param use_conv: a bool determining if a convolution is applied.
+    :param dims: determines if the signal is 1D, 2D, or 3D. If 3D, then
+                 downsampling occurs in the inner-two dimensions.
+    """
+
+    def __init__(self, channels, use_conv, dims=2, out_channels=None, padding=1, dtype=None, device=None, operations=fcbh.ops):
+        super().__init__()
+        self.channels = channels
+        self.out_channels = out_channels or channels
+        self.use_conv = use_conv
+        self.dims = dims
+        stride = 2 if dims != 3 else (1, 2, 2)
+        if use_conv:
+            self.op = operations.conv_nd(
+                dims, self.channels, self.out_channels, 3, stride=stride, padding=padding, dtype=dtype, device=device
+            )
+        else:
+            assert self.channels == self.out_channels
+            self.op = avg_pool_nd(dims, kernel_size=stride, stride=stride)
+
+    def forward(self, x):
+        assert x.shape[1] == self.channels
+        return self.op(x)
+
+
+class ResBlock(TimestepBlock):
+    """
+    A residual block that can optionally change the number of channels.
+    :param channels: the number of input channels.
+    :param emb_channels: the number of timestep embedding channels.
+    :param dropout: the rate of dropout.
+    :param out_channels: if specified, the number of out channels.
+    :param use_conv: if True and out_channels is specified, use a spatial
+        convolution instead of a smaller 1x1 convolution to change the
+        channels in the skip connection.
+    :param dims: determines if the signal is 1D, 2D, or 3D.
+    :param use_checkpoint: if True, use gradient checkpointing on this module.
+    :param up: if True, use this block for upsampling.
+    :param down: if True, use this block for downsampling.
+    """
+
+    def __init__(
+        self,
+        channels,
+        emb_channels,
+        dropout,
+        out_channels=None,
+        use_conv=False,
+        use_scale_shift_norm=False,
+        dims=2,
+        use_checkpoint=False,
+        up=False,
+        down=False,
+        dtype=None,
+        device=None,
+        operations=fcbh.ops
+    ):
+        super().__init__()
+        self.channels = channels
+        self.emb_channels = emb_channels
+        self.dropout = dropout
+        self.out_channels = out_channels or channels
+        self.use_conv = use_conv
+        self.use_checkpoint = use_checkpoint
+        self.use_scale_shift_norm = use_scale_shift_norm
+
+        self.in_layers = nn.Sequential(
+            nn.GroupNorm(32, channels, dtype=dtype, device=device),
+            nn.SiLU(),
+            operations.conv_nd(dims, channels, self.out_channels, 3, padding=1, dtype=dtype, device=device),
+        )
+
+        self.updown = up or down
+
+        if up:
+            self.h_upd = Upsample(channels, False, dims, dtype=dtype, device=device)
+            self.x_upd = Upsample(channels, False, dims, dtype=dtype, device=device)
+        elif down:
+            self.h_upd = Downsample(channels, False, dims, dtype=dtype, device=device)
+            self.x_upd = Downsample(channels, False, dims, dtype=dtype, device=device)
+        else:
+            self.h_upd = self.x_upd = nn.Identity()
+
+        self.emb_layers = nn.Sequential(
+            nn.SiLU(),
+            operations.Linear(
+                emb_channels,
+                2 * self.out_channels if use_scale_shift_norm else self.out_channels, dtype=dtype, device=device
+            ),
+        )
+        self.out_layers = nn.Sequential(
+            nn.GroupNorm(32, self.out_channels, dtype=dtype, device=device),
+            nn.SiLU(),
+            nn.Dropout(p=dropout),
+            zero_module(
+                operations.conv_nd(dims, self.out_channels, self.out_channels, 3, padding=1, dtype=dtype, device=device)
+            ),
+        )
+
+        if self.out_channels == channels:
+            self.skip_connection = nn.Identity()
+        elif use_conv:
+            self.skip_connection = operations.conv_nd(
+                dims, channels, self.out_channels, 3, padding=1, dtype=dtype, device=device
+            )
+        else:
+            self.skip_connection = operations.conv_nd(dims, channels, self.out_channels, 1, dtype=dtype, device=device)
+
+    def forward(self, x, emb):
+        """
+        Apply the block to a Tensor, conditioned on a timestep embedding.
+        :param x: an [N x C x ...] Tensor of features.
+        :param emb: an [N x emb_channels] Tensor of timestep embeddings.
+        :return: an [N x C x ...] Tensor of outputs.
+        """
+        return checkpoint(
+            self._forward, (x, emb), self.parameters(), self.use_checkpoint
+        )
+
+
+    def _forward(self, x, emb):
+        if self.updown:
+            in_rest, in_conv = self.in_layers[:-1], self.in_layers[-1]
+            h = in_rest(x)
+            h = self.h_upd(h)
+            x = self.x_upd(x)
+            h = in_conv(h)
+        else:
+            h = self.in_layers(x)
+        emb_out = self.emb_layers(emb).type(h.dtype)
+        while len(emb_out.shape) < len(h.shape):
+            emb_out = emb_out[..., None]
+        if self.use_scale_shift_norm:
+            out_norm, out_rest = self.out_layers[0], self.out_layers[1:]
+            scale, shift = th.chunk(emb_out, 2, dim=1)
+            h = out_norm(h) * (1 + scale) + shift
+            h = out_rest(h)
+        else:
+            h = h + emb_out
+            h = self.out_layers(h)
+        return self.skip_connection(x) + h
+
+class Timestep(nn.Module):
+    def __init__(self, dim):
+        super().__init__()
+        self.dim = dim
+
+    def forward(self, t):
+        return timestep_embedding(t, self.dim)
+
+def apply_control(h, control, name):
+    if control is not None and name in control and len(control[name]) > 0:
+        ctrl = control[name].pop()
+        if ctrl is not None:
+            h += ctrl
+    return h
+
+class UNetModel(nn.Module):
+    """
+    The full UNet model with attention and timestep embedding.
+    :param in_channels: channels in the input Tensor.
+    :param model_channels: base channel count for the model.
+    :param out_channels: channels in the output Tensor.
+    :param num_res_blocks: number of residual blocks per downsample.
+    :param dropout: the dropout probability.
+    :param channel_mult: channel multiplier for each level of the UNet.
+    :param conv_resample: if True, use learned convolutions for upsampling and
+        downsampling.
+    :param dims: determines if the signal is 1D, 2D, or 3D.
+    :param num_classes: if specified (as an int), then this model will be
+        class-conditional with `num_classes` classes.
+    :param use_checkpoint: use gradient checkpointing to reduce memory usage.
+    :param num_heads: the number of attention heads in each attention layer.
+    :param num_heads_channels: if specified, ignore num_heads and instead use
+                               a fixed channel width per attention head.
+    :param num_heads_upsample: works with num_heads to set a different number
+                               of heads for upsampling. Deprecated.
+    :param use_scale_shift_norm: use a FiLM-like conditioning mechanism.
+    :param resblock_updown: use residual blocks for up/downsampling.
+    :param use_new_attention_order: use a different attention pattern for potentially
+                                    increased efficiency.
+    """
+
+    def __init__(
+        self,
+        image_size,
+        in_channels,
+        model_channels,
+        out_channels,
+        num_res_blocks,
+        dropout=0,
+        channel_mult=(1, 2, 4, 8),
+        conv_resample=True,
+        dims=2,
+        num_classes=None,
+        use_checkpoint=False,
+        dtype=th.float32,
+        num_heads=-1,
+        num_head_channels=-1,
+        num_heads_upsample=-1,
+        use_scale_shift_norm=False,
+        resblock_updown=False,
+        use_new_attention_order=False,
+        use_spatial_transformer=False,    # custom transformer support
+        transformer_depth=1,              # custom transformer support
+        context_dim=None,                 # custom transformer support
+        n_embed=None,                     # custom support for prediction of discrete ids into codebook of first stage vq model
+        legacy=True,
+        disable_self_attentions=None,
+        num_attention_blocks=None,
+        disable_middle_self_attn=False,
+        use_linear_in_transformer=False,
+        adm_in_channels=None,
+        transformer_depth_middle=None,
+        transformer_depth_output=None,
+        device=None,
+        operations=fcbh.ops,
+    ):
+        super().__init__()
+        assert use_spatial_transformer == True, "use_spatial_transformer has to be true"
+        if use_spatial_transformer:
+            assert context_dim is not None, 'Fool!! You forgot to include the dimension of your cross-attention conditioning...'
+
+        if context_dim is not None:
+            assert use_spatial_transformer, 'Fool!! You forgot to use the spatial transformer for your cross-attention conditioning...'
+            # from omegaconf.listconfig import ListConfig
+            # if type(context_dim) == ListConfig:
+            #     context_dim = list(context_dim)
+
+        if num_heads_upsample == -1:
+            num_heads_upsample = num_heads
+
+        if num_heads == -1:
+            assert num_head_channels != -1, 'Either num_heads or num_head_channels has to be set'
+
+        if num_head_channels == -1:
+            assert num_heads != -1, 'Either num_heads or num_head_channels has to be set'
+
+        self.image_size = image_size
+        self.in_channels = in_channels
+        self.model_channels = model_channels
+        self.out_channels = out_channels
+
+        if isinstance(num_res_blocks, int):
+            self.num_res_blocks = len(channel_mult) * [num_res_blocks]
+        else:
+            if len(num_res_blocks) != len(channel_mult):
+                raise ValueError("provide num_res_blocks either as an int (globally constant) or "
+                                 "as a list/tuple (per-level) with the same length as channel_mult")
+            self.num_res_blocks = num_res_blocks
+
+        if disable_self_attentions is not None:
+            # should be a list of booleans, indicating whether to disable self-attention in TransformerBlocks or not
+            assert len(disable_self_attentions) == len(channel_mult)
+        if num_attention_blocks is not None:
+            assert len(num_attention_blocks) == len(self.num_res_blocks)
+
+        transformer_depth = transformer_depth[:]
+        transformer_depth_output = transformer_depth_output[:]
+
+        self.dropout = dropout
+        self.channel_mult = channel_mult
+        self.conv_resample = conv_resample
+        self.num_classes = num_classes
+        self.use_checkpoint = use_checkpoint
+        self.dtype = dtype
+        self.num_heads = num_heads
+        self.num_head_channels = num_head_channels
+        self.num_heads_upsample = num_heads_upsample
+        self.predict_codebook_ids = n_embed is not None
+
+        time_embed_dim = model_channels * 4
+        self.time_embed = nn.Sequential(
+            operations.Linear(model_channels, time_embed_dim, dtype=self.dtype, device=device),
+            nn.SiLU(),
+            operations.Linear(time_embed_dim, time_embed_dim, dtype=self.dtype, device=device),
+        )
+
+        if self.num_classes is not None:
+            if isinstance(self.num_classes, int):
+                self.label_emb = nn.Embedding(num_classes, time_embed_dim)
+            elif self.num_classes == "continuous":
+                print("setting up linear c_adm embedding layer")
+                self.label_emb = nn.Linear(1, time_embed_dim)
+            elif self.num_classes == "sequential":
+                assert adm_in_channels is not None
+                self.label_emb = nn.Sequential(
+                    nn.Sequential(
+                        operations.Linear(adm_in_channels, time_embed_dim, dtype=self.dtype, device=device),
+                        nn.SiLU(),
+                        operations.Linear(time_embed_dim, time_embed_dim, dtype=self.dtype, device=device),
+                    )
+                )
+            else:
+                raise ValueError()
+
+        self.input_blocks = nn.ModuleList(
+            [
+                TimestepEmbedSequential(
+                    operations.conv_nd(dims, in_channels, model_channels, 3, padding=1, dtype=self.dtype, device=device)
+                )
+            ]
+        )
+        self._feature_size = model_channels
+        input_block_chans = [model_channels]
+        ch = model_channels
+        ds = 1
+        for level, mult in enumerate(channel_mult):
+            for nr in range(self.num_res_blocks[level]):
+                layers = [
+                    ResBlock(
+                        ch,
+                        time_embed_dim,
+                        dropout,
+                        out_channels=mult * model_channels,
+                        dims=dims,
+                        use_checkpoint=use_checkpoint,
+                        use_scale_shift_norm=use_scale_shift_norm,
+                        dtype=self.dtype,
+                        device=device,
+                        operations=operations,
+                    )
+                ]
+                ch = mult * model_channels
+                num_transformers = transformer_depth.pop(0)
+                if num_transformers > 0:
+                    if num_head_channels == -1:
+                        dim_head = ch // num_heads
+                    else:
+                        num_heads = ch // num_head_channels
+                        dim_head = num_head_channels
+                    if legacy:
+                        #num_heads = 1
+                        dim_head = ch // num_heads if use_spatial_transformer else num_head_channels
+                    if exists(disable_self_attentions):
+                        disabled_sa = disable_self_attentions[level]
+                    else:
+                        disabled_sa = False
+
+                    if not exists(num_attention_blocks) or nr < num_attention_blocks[level]:
+                        layers.append(SpatialTransformer(
+                                ch, num_heads, dim_head, depth=num_transformers, context_dim=context_dim,
+                                disable_self_attn=disabled_sa, use_linear=use_linear_in_transformer,
+                                use_checkpoint=use_checkpoint, dtype=self.dtype, device=device, operations=operations
+                            )
+                        )
+                self.input_blocks.append(TimestepEmbedSequential(*layers))
+                self._feature_size += ch
+                input_block_chans.append(ch)
+            if level != len(channel_mult) - 1:
+                out_ch = ch
+                self.input_blocks.append(
+                    TimestepEmbedSequential(
+                        ResBlock(
+                            ch,
+                            time_embed_dim,
+                            dropout,
+                            out_channels=out_ch,
+                            dims=dims,
+                            use_checkpoint=use_checkpoint,
+                            use_scale_shift_norm=use_scale_shift_norm,
+                            down=True,
+                            dtype=self.dtype,
+                            device=device,
+                            operations=operations
+                        )
+                        if resblock_updown
+                        else Downsample(
+                            ch, conv_resample, dims=dims, out_channels=out_ch, dtype=self.dtype, device=device, operations=operations
+                        )
+                    )
+                )
+                ch = out_ch
+                input_block_chans.append(ch)
+                ds *= 2
+                self._feature_size += ch
+
+        if num_head_channels == -1:
+            dim_head = ch // num_heads
+        else:
+            num_heads = ch // num_head_channels
+            dim_head = num_head_channels
+        if legacy:
+            #num_heads = 1
+            dim_head = ch // num_heads if use_spatial_transformer else num_head_channels
+        mid_block = [
+            ResBlock(
+                ch,
+                time_embed_dim,
+                dropout,
+                dims=dims,
+                use_checkpoint=use_checkpoint,
+                use_scale_shift_norm=use_scale_shift_norm,
+                dtype=self.dtype,
+                device=device,
+                operations=operations
+            )]
+        if transformer_depth_middle >= 0:
+            mid_block += [SpatialTransformer(  # always uses a self-attn
+                            ch, num_heads, dim_head, depth=transformer_depth_middle, context_dim=context_dim,
+                            disable_self_attn=disable_middle_self_attn, use_linear=use_linear_in_transformer,
+                            use_checkpoint=use_checkpoint, dtype=self.dtype, device=device, operations=operations
+                        ),
+            ResBlock(
+                ch,
+                time_embed_dim,
+                dropout,
+                dims=dims,
+                use_checkpoint=use_checkpoint,
+                use_scale_shift_norm=use_scale_shift_norm,
+                dtype=self.dtype,
+                device=device,
+                operations=operations
+            )]
+        self.middle_block = TimestepEmbedSequential(*mid_block)
+        self._feature_size += ch
+
+        self.output_blocks = nn.ModuleList([])
+        for level, mult in list(enumerate(channel_mult))[::-1]:
+            for i in range(self.num_res_blocks[level] + 1):
+                ich = input_block_chans.pop()
+                layers = [
+                    ResBlock(
+                        ch + ich,
+                        time_embed_dim,
+                        dropout,
+                        out_channels=model_channels * mult,
+                        dims=dims,
+                        use_checkpoint=use_checkpoint,
+                        use_scale_shift_norm=use_scale_shift_norm,
+                        dtype=self.dtype,
+                        device=device,
+                        operations=operations
+                    )
+                ]
+                ch = model_channels * mult
+                num_transformers = transformer_depth_output.pop()
+                if num_transformers > 0:
+                    if num_head_channels == -1:
+                        dim_head = ch // num_heads
+                    else:
+                        num_heads = ch // num_head_channels
+                        dim_head = num_head_channels
+                    if legacy:
+                        #num_heads = 1
+                        dim_head = ch // num_heads if use_spatial_transformer else num_head_channels
+                    if exists(disable_self_attentions):
+                        disabled_sa = disable_self_attentions[level]
+                    else:
+                        disabled_sa = False
+
+                    if not exists(num_attention_blocks) or i < num_attention_blocks[level]:
+                        layers.append(
+                            SpatialTransformer(
+                                ch, num_heads, dim_head, depth=num_transformers, context_dim=context_dim,
+                                disable_self_attn=disabled_sa, use_linear=use_linear_in_transformer,
+                                use_checkpoint=use_checkpoint, dtype=self.dtype, device=device, operations=operations
+                            )
+                        )
+                if level and i == self.num_res_blocks[level]:
+                    out_ch = ch
+                    layers.append(
+                        ResBlock(
+                            ch,
+                            time_embed_dim,
+                            dropout,
+                            out_channels=out_ch,
+                            dims=dims,
+                            use_checkpoint=use_checkpoint,
+                            use_scale_shift_norm=use_scale_shift_norm,
+                            up=True,
+                            dtype=self.dtype,
+                            device=device,
+                            operations=operations
+                        )
+                        if resblock_updown
+                        else Upsample(ch, conv_resample, dims=dims, out_channels=out_ch, dtype=self.dtype, device=device, operations=operations)
+                    )
+                    ds //= 2
+                self.output_blocks.append(TimestepEmbedSequential(*layers))
+                self._feature_size += ch
+
+        self.out = nn.Sequential(
+            nn.GroupNorm(32, ch, dtype=self.dtype, device=device),
+            nn.SiLU(),
+            zero_module(operations.conv_nd(dims, model_channels, out_channels, 3, padding=1, dtype=self.dtype, device=device)),
+        )
+        if self.predict_codebook_ids:
+            self.id_predictor = nn.Sequential(
+            nn.GroupNorm(32, ch, dtype=self.dtype, device=device),
+            operations.conv_nd(dims, model_channels, n_embed, 1, dtype=self.dtype, device=device),
+            #nn.LogSoftmax(dim=1)  # change to cross_entropy and produce non-normalized logits
+        )
+
+    def forward(self, x, timesteps=None, context=None, y=None, control=None, transformer_options={}, **kwargs):
+        """
+        Apply the model to an input batch.
+        :param x: an [N x C x ...] Tensor of inputs.
+        :param timesteps: a 1-D batch of timesteps.
+        :param context: conditioning plugged in via crossattn
+        :param y: an [N] Tensor of labels, if class-conditional.
+        :return: an [N x C x ...] Tensor of outputs.
+        """
+        transformer_options["original_shape"] = list(x.shape)
+        transformer_options["current_index"] = 0
+        transformer_patches = transformer_options.get("patches", {})
+
+        assert (y is not None) == (
+            self.num_classes is not None
+        ), "must specify y if and only if the model is class-conditional"
+        hs = []
+        t_emb = timestep_embedding(timesteps, self.model_channels, repeat_only=False).to(self.dtype)
+        emb = self.time_embed(t_emb)
+
+        if self.num_classes is not None:
+            assert y.shape[0] == x.shape[0]
+            emb = emb + self.label_emb(y)
+
+        h = x.type(self.dtype)
+        for id, module in enumerate(self.input_blocks):
+            transformer_options["block"] = ("input", id)
+            h = forward_timestep_embed(module, h, emb, context, transformer_options)
+            h = apply_control(h, control, 'input')
+            hs.append(h)
+
+        transformer_options["block"] = ("middle", 0)
+        h = forward_timestep_embed(self.middle_block, h, emb, context, transformer_options)
+        h = apply_control(h, control, 'middle')
+
+        for id, module in enumerate(self.output_blocks):
+            transformer_options["block"] = ("output", id)
+            hsp = hs.pop()
+            hsp = apply_control(hsp, control, 'output')
+
+            if "output_block_patch" in transformer_patches:
+                patch = transformer_patches["output_block_patch"]
+                for p in patch:
+                    h, hsp = p(h, hsp, transformer_options)
+
+            h = th.cat([h, hsp], dim=1)
+            del hsp
+            if len(hs) > 0:
+                output_shape = hs[-1].shape
+            else:
+                output_shape = None
+            h = forward_timestep_embed(module, h, emb, context, transformer_options, output_shape)
+        h = h.type(x.dtype)
+        if self.predict_codebook_ids:
+            return self.id_predictor(h)
+        else:
+            return self.out(h)

backend/headless/fcbh/ldm/modules/diffusionmodules/upscaling.py

@@ -0,0 +1,81 @@
+import torch
+import torch.nn as nn
+import numpy as np
+from functools import partial
+
+from .util import extract_into_tensor, make_beta_schedule
+from fcbh.ldm.util import default
+
+
+class AbstractLowScaleModel(nn.Module):
+    # for concatenating a downsampled image to the latent representation
+    def __init__(self, noise_schedule_config=None):
+        super(AbstractLowScaleModel, self).__init__()
+        if noise_schedule_config is not None:
+            self.register_schedule(**noise_schedule_config)
+
+    def register_schedule(self, beta_schedule="linear", timesteps=1000,
+                          linear_start=1e-4, linear_end=2e-2, cosine_s=8e-3):
+        betas = make_beta_schedule(beta_schedule, timesteps, linear_start=linear_start, linear_end=linear_end,
+                                   cosine_s=cosine_s)
+        alphas = 1. - betas
+        alphas_cumprod = np.cumprod(alphas, axis=0)
+        alphas_cumprod_prev = np.append(1., alphas_cumprod[:-1])
+
+        timesteps, = betas.shape
+        self.num_timesteps = int(timesteps)
+        self.linear_start = linear_start
+        self.linear_end = linear_end
+        assert alphas_cumprod.shape[0] == self.num_timesteps, 'alphas have to be defined for each timestep'
+
+        to_torch = partial(torch.tensor, dtype=torch.float32)
+
+        self.register_buffer('betas', to_torch(betas))
+        self.register_buffer('alphas_cumprod', to_torch(alphas_cumprod))
+        self.register_buffer('alphas_cumprod_prev', to_torch(alphas_cumprod_prev))
+
+        # calculations for diffusion q(x_t | x_{t-1}) and others
+        self.register_buffer('sqrt_alphas_cumprod', to_torch(np.sqrt(alphas_cumprod)))
+        self.register_buffer('sqrt_one_minus_alphas_cumprod', to_torch(np.sqrt(1. - alphas_cumprod)))
+        self.register_buffer('log_one_minus_alphas_cumprod', to_torch(np.log(1. - alphas_cumprod)))
+        self.register_buffer('sqrt_recip_alphas_cumprod', to_torch(np.sqrt(1. / alphas_cumprod)))
+        self.register_buffer('sqrt_recipm1_alphas_cumprod', to_torch(np.sqrt(1. / alphas_cumprod - 1)))
+
+    def q_sample(self, x_start, t, noise=None):
+        noise = default(noise, lambda: torch.randn_like(x_start))
+        return (extract_into_tensor(self.sqrt_alphas_cumprod, t, x_start.shape) * x_start +
+                extract_into_tensor(self.sqrt_one_minus_alphas_cumprod, t, x_start.shape) * noise)
+
+    def forward(self, x):
+        return x, None
+
+    def decode(self, x):
+        return x
+
+
+class SimpleImageConcat(AbstractLowScaleModel):
+    # no noise level conditioning
+    def __init__(self):
+        super(SimpleImageConcat, self).__init__(noise_schedule_config=None)
+        self.max_noise_level = 0
+
+    def forward(self, x):
+        # fix to constant noise level
+        return x, torch.zeros(x.shape[0], device=x.device).long()
+
+
+class ImageConcatWithNoiseAugmentation(AbstractLowScaleModel):
+    def __init__(self, noise_schedule_config, max_noise_level=1000, to_cuda=False):
+        super().__init__(noise_schedule_config=noise_schedule_config)
+        self.max_noise_level = max_noise_level
+
+    def forward(self, x, noise_level=None):
+        if noise_level is None:
+            noise_level = torch.randint(0, self.max_noise_level, (x.shape[0],), device=x.device).long()
+        else:
+            assert isinstance(noise_level, torch.Tensor)
+        z = self.q_sample(x, noise_level)
+        return z, noise_level
+
+
+

backend/headless/fcbh/ldm/modules/diffusionmodules/util.py

@@ -0,0 +1,278 @@
+# adopted from
+# https://github.com/openai/improved-diffusion/blob/main/improved_diffusion/gaussian_diffusion.py
+# and
+# https://github.com/lucidrains/denoising-diffusion-pytorch/blob/7706bdfc6f527f58d33f84b7b522e61e6e3164b3/denoising_diffusion_pytorch/denoising_diffusion_pytorch.py
+# and
+# https://github.com/openai/guided-diffusion/blob/0ba878e517b276c45d1195eb29f6f5f72659a05b/guided_diffusion/nn.py
+#
+# thanks!
+
+
+import os
+import math
+import torch
+import torch.nn as nn
+import numpy as np
+from einops import repeat
+
+from fcbh.ldm.util import instantiate_from_config
+import fcbh.ops
+
+def make_beta_schedule(schedule, n_timestep, linear_start=1e-4, linear_end=2e-2, cosine_s=8e-3):
+    if schedule == "linear":
+        betas = (
+                torch.linspace(linear_start ** 0.5, linear_end ** 0.5, n_timestep, dtype=torch.float64) ** 2
+        )
+
+    elif schedule == "cosine":
+        timesteps = (
+                torch.arange(n_timestep + 1, dtype=torch.float64) / n_timestep + cosine_s
+        )
+        alphas = timesteps / (1 + cosine_s) * np.pi / 2
+        alphas = torch.cos(alphas).pow(2)
+        alphas = alphas / alphas[0]
+        betas = 1 - alphas[1:] / alphas[:-1]
+        betas = np.clip(betas, a_min=0, a_max=0.999)
+
+    elif schedule == "squaredcos_cap_v2":  # used for karlo prior
+        # return early
+        return betas_for_alpha_bar(
+            n_timestep,
+            lambda t: math.cos((t + 0.008) / 1.008 * math.pi / 2) ** 2,
+        )
+
+    elif schedule == "sqrt_linear":
+        betas = torch.linspace(linear_start, linear_end, n_timestep, dtype=torch.float64)
+    elif schedule == "sqrt":
+        betas = torch.linspace(linear_start, linear_end, n_timestep, dtype=torch.float64) ** 0.5
+    else:
+        raise ValueError(f"schedule '{schedule}' unknown.")
+    return betas.numpy()
+
+
+def make_ddim_timesteps(ddim_discr_method, num_ddim_timesteps, num_ddpm_timesteps, verbose=True):
+    if ddim_discr_method == 'uniform':
+        c = num_ddpm_timesteps // num_ddim_timesteps
+        ddim_timesteps = np.asarray(list(range(0, num_ddpm_timesteps, c)))
+    elif ddim_discr_method == 'quad':
+        ddim_timesteps = ((np.linspace(0, np.sqrt(num_ddpm_timesteps * .8), num_ddim_timesteps)) ** 2).astype(int)
+    else:
+        raise NotImplementedError(f'There is no ddim discretization method called "{ddim_discr_method}"')
+
+    # assert ddim_timesteps.shape[0] == num_ddim_timesteps
+    # add one to get the final alpha values right (the ones from first scale to data during sampling)
+    steps_out = ddim_timesteps + 1
+    if verbose:
+        print(f'Selected timesteps for ddim sampler: {steps_out}')
+    return steps_out
+
+
+def make_ddim_sampling_parameters(alphacums, ddim_timesteps, eta, verbose=True):
+    # select alphas for computing the variance schedule
+    alphas = alphacums[ddim_timesteps]
+    alphas_prev = np.asarray([alphacums[0]] + alphacums[ddim_timesteps[:-1]].tolist())
+
+    # according the the formula provided in https://arxiv.org/abs/2010.02502
+    sigmas = eta * np.sqrt((1 - alphas_prev) / (1 - alphas) * (1 - alphas / alphas_prev))
+    if verbose:
+        print(f'Selected alphas for ddim sampler: a_t: {alphas}; a_(t-1): {alphas_prev}')
+        print(f'For the chosen value of eta, which is {eta}, '
+              f'this results in the following sigma_t schedule for ddim sampler {sigmas}')
+    return sigmas, alphas, alphas_prev
+
+
+def betas_for_alpha_bar(num_diffusion_timesteps, alpha_bar, max_beta=0.999):
+    """
+    Create a beta schedule that discretizes the given alpha_t_bar function,
+    which defines the cumulative product of (1-beta) over time from t = [0,1].
+    :param num_diffusion_timesteps: the number of betas to produce.
+    :param alpha_bar: a lambda that takes an argument t from 0 to 1 and
+                      produces the cumulative product of (1-beta) up to that
+                      part of the diffusion process.
+    :param max_beta: the maximum beta to use; use values lower than 1 to
+                     prevent singularities.
+    """
+    betas = []
+    for i in range(num_diffusion_timesteps):
+        t1 = i / num_diffusion_timesteps
+        t2 = (i + 1) / num_diffusion_timesteps
+        betas.append(min(1 - alpha_bar(t2) / alpha_bar(t1), max_beta))
+    return np.array(betas)
+
+
+def extract_into_tensor(a, t, x_shape):
+    b, *_ = t.shape
+    out = a.gather(-1, t)
+    return out.reshape(b, *((1,) * (len(x_shape) - 1)))
+
+
+def checkpoint(func, inputs, params, flag):
+    """
+    Evaluate a function without caching intermediate activations, allowing for
+    reduced memory at the expense of extra compute in the backward pass.
+    :param func: the function to evaluate.
+    :param inputs: the argument sequence to pass to `func`.
+    :param params: a sequence of parameters `func` depends on but does not
+                   explicitly take as arguments.
+    :param flag: if False, disable gradient checkpointing.
+    """
+    if flag:
+        args = tuple(inputs) + tuple(params)
+        return CheckpointFunction.apply(func, len(inputs), *args)
+    else:
+        return func(*inputs)
+
+
+class CheckpointFunction(torch.autograd.Function):
+    @staticmethod
+    def forward(ctx, run_function, length, *args):
+        ctx.run_function = run_function
+        ctx.input_tensors = list(args[:length])
+        ctx.input_params = list(args[length:])
+        ctx.gpu_autocast_kwargs = {"enabled": torch.is_autocast_enabled(),
+                                   "dtype": torch.get_autocast_gpu_dtype(),
+                                   "cache_enabled": torch.is_autocast_cache_enabled()}
+        with torch.no_grad():
+            output_tensors = ctx.run_function(*ctx.input_tensors)
+        return output_tensors
+
+    @staticmethod
+    def backward(ctx, *output_grads):
+        ctx.input_tensors = [x.detach().requires_grad_(True) for x in ctx.input_tensors]
+        with torch.enable_grad(), \
+                torch.cuda.amp.autocast(**ctx.gpu_autocast_kwargs):
+            # Fixes a bug where the first op in run_function modifies the
+            # Tensor storage in place, which is not allowed for detach()'d
+            # Tensors.
+            shallow_copies = [x.view_as(x) for x in ctx.input_tensors]
+            output_tensors = ctx.run_function(*shallow_copies)
+        input_grads = torch.autograd.grad(
+            output_tensors,
+            ctx.input_tensors + ctx.input_params,
+            output_grads,
+            allow_unused=True,
+        )
+        del ctx.input_tensors
+        del ctx.input_params
+        del output_tensors
+        return (None, None) + input_grads
+
+
+def timestep_embedding(timesteps, dim, max_period=10000, repeat_only=False):
+    """
+    Create sinusoidal timestep embeddings.
+    :param timesteps: a 1-D Tensor of N indices, one per batch element.
+                      These may be fractional.
+    :param dim: the dimension of the output.
+    :param max_period: controls the minimum frequency of the embeddings.
+    :return: an [N x dim] Tensor of positional embeddings.
+    """
+    if not repeat_only:
+        half = dim // 2
+        freqs = torch.exp(
+            -math.log(max_period) * torch.arange(start=0, end=half, dtype=torch.float32, device=timesteps.device) / half
+        )
+        args = timesteps[:, None].float() * freqs[None]
+        embedding = torch.cat([torch.cos(args), torch.sin(args)], dim=-1)
+        if dim % 2:
+            embedding = torch.cat([embedding, torch.zeros_like(embedding[:, :1])], dim=-1)
+    else:
+        embedding = repeat(timesteps, 'b -> b d', d=dim)
+    return embedding
+
+
+def zero_module(module):
+    """
+    Zero out the parameters of a module and return it.
+    """
+    for p in module.parameters():
+        p.detach().zero_()
+    return module
+
+
+def scale_module(module, scale):
+    """
+    Scale the parameters of a module and return it.
+    """
+    for p in module.parameters():
+        p.detach().mul_(scale)
+    return module
+
+
+def mean_flat(tensor):
+    """
+    Take the mean over all non-batch dimensions.
+    """
+    return tensor.mean(dim=list(range(1, len(tensor.shape))))
+
+
+def normalization(channels, dtype=None):
+    """
+    Make a standard normalization layer.
+    :param channels: number of input channels.
+    :return: an nn.Module for normalization.
+    """
+    return GroupNorm32(32, channels, dtype=dtype)
+
+
+# PyTorch 1.7 has SiLU, but we support PyTorch 1.5.
+class SiLU(nn.Module):
+    def forward(self, x):
+        return x * torch.sigmoid(x)
+
+
+class GroupNorm32(nn.GroupNorm):
+    def forward(self, x):
+        return super().forward(x.float()).type(x.dtype)
+
+
+def conv_nd(dims, *args, **kwargs):
+    """
+    Create a 1D, 2D, or 3D convolution module.
+    """
+    if dims == 1:
+        return nn.Conv1d(*args, **kwargs)
+    elif dims == 2:
+        return fcbh.ops.Conv2d(*args, **kwargs)
+    elif dims == 3:
+        return nn.Conv3d(*args, **kwargs)
+    raise ValueError(f"unsupported dimensions: {dims}")
+
+
+def linear(*args, **kwargs):
+    """
+    Create a linear module.
+    """
+    return fcbh.ops.Linear(*args, **kwargs)
+
+
+def avg_pool_nd(dims, *args, **kwargs):
+    """
+    Create a 1D, 2D, or 3D average pooling module.
+    """
+    if dims == 1:
+        return nn.AvgPool1d(*args, **kwargs)
+    elif dims == 2:
+        return nn.AvgPool2d(*args, **kwargs)
+    elif dims == 3:
+        return nn.AvgPool3d(*args, **kwargs)
+    raise ValueError(f"unsupported dimensions: {dims}")
+
+
+class HybridConditioner(nn.Module):
+
+    def __init__(self, c_concat_config, c_crossattn_config):
+        super().__init__()
+        self.concat_conditioner = instantiate_from_config(c_concat_config)
+        self.crossattn_conditioner = instantiate_from_config(c_crossattn_config)
+
+    def forward(self, c_concat, c_crossattn):
+        c_concat = self.concat_conditioner(c_concat)
+        c_crossattn = self.crossattn_conditioner(c_crossattn)
+        return {'c_concat': [c_concat], 'c_crossattn': [c_crossattn]}
+
+
+def noise_like(shape, device, repeat=False):
+    repeat_noise = lambda: torch.randn((1, *shape[1:]), device=device).repeat(shape[0], *((1,) * (len(shape) - 1)))
+    noise = lambda: torch.randn(shape, device=device)
+    return repeat_noise() if repeat else noise()

backend/headless/fcbh/ldm/modules/distributions/distributions.py

@@ -0,0 +1,92 @@
+import torch
+import numpy as np
+
+
+class AbstractDistribution:
+    def sample(self):
+        raise NotImplementedError()
+
+    def mode(self):
+        raise NotImplementedError()
+
+
+class DiracDistribution(AbstractDistribution):
+    def __init__(self, value):
+        self.value = value
+
+    def sample(self):
+        return self.value
+
+    def mode(self):
+        return self.value
+
+
+class DiagonalGaussianDistribution(object):
+    def __init__(self, parameters, deterministic=False):
+        self.parameters = parameters
+        self.mean, self.logvar = torch.chunk(parameters, 2, dim=1)
+        self.logvar = torch.clamp(self.logvar, -30.0, 20.0)
+        self.deterministic = deterministic
+        self.std = torch.exp(0.5 * self.logvar)
+        self.var = torch.exp(self.logvar)
+        if self.deterministic:
+            self.var = self.std = torch.zeros_like(self.mean).to(device=self.parameters.device)
+
+    def sample(self):
+        x = self.mean + self.std * torch.randn(self.mean.shape).to(device=self.parameters.device)
+        return x
+
+    def kl(self, other=None):
+        if self.deterministic:
+            return torch.Tensor([0.])
+        else:
+            if other is None:
+                return 0.5 * torch.sum(torch.pow(self.mean, 2)
+                                       + self.var - 1.0 - self.logvar,
+                                       dim=[1, 2, 3])
+            else:
+                return 0.5 * torch.sum(
+                    torch.pow(self.mean - other.mean, 2) / other.var
+                    + self.var / other.var - 1.0 - self.logvar + other.logvar,
+                    dim=[1, 2, 3])
+
+    def nll(self, sample, dims=[1,2,3]):
+        if self.deterministic:
+            return torch.Tensor([0.])
+        logtwopi = np.log(2.0 * np.pi)
+        return 0.5 * torch.sum(
+            logtwopi + self.logvar + torch.pow(sample - self.mean, 2) / self.var,
+            dim=dims)
+
+    def mode(self):
+        return self.mean
+
+
+def normal_kl(mean1, logvar1, mean2, logvar2):
+    """
+    source: https://github.com/openai/guided-diffusion/blob/27c20a8fab9cb472df5d6bdd6c8d11c8f430b924/guided_diffusion/losses.py#L12
+    Compute the KL divergence between two gaussians.
+    Shapes are automatically broadcasted, so batches can be compared to
+    scalars, among other use cases.
+    """
+    tensor = None
+    for obj in (mean1, logvar1, mean2, logvar2):
+        if isinstance(obj, torch.Tensor):
+            tensor = obj
+            break
+    assert tensor is not None, "at least one argument must be a Tensor"
+
+    # Force variances to be Tensors. Broadcasting helps convert scalars to
+    # Tensors, but it does not work for torch.exp().
+    logvar1, logvar2 = [
+        x if isinstance(x, torch.Tensor) else torch.tensor(x).to(tensor)
+        for x in (logvar1, logvar2)
+    ]
+
+    return 0.5 * (
+        -1.0
+        + logvar2
+        - logvar1
+        + torch.exp(logvar1 - logvar2)
+        + ((mean1 - mean2) ** 2) * torch.exp(-logvar2)
+    )

backend/headless/fcbh/ldm/modules/ema.py

@@ -0,0 +1,80 @@
+import torch
+from torch import nn
+
+
+class LitEma(nn.Module):
+    def __init__(self, model, decay=0.9999, use_num_upates=True):
+        super().__init__()
+        if decay < 0.0 or decay > 1.0:
+            raise ValueError('Decay must be between 0 and 1')
+
+        self.m_name2s_name = {}
+        self.register_buffer('decay', torch.tensor(decay, dtype=torch.float32))
+        self.register_buffer('num_updates', torch.tensor(0, dtype=torch.int) if use_num_upates
+        else torch.tensor(-1, dtype=torch.int))
+
+        for name, p in model.named_parameters():
+            if p.requires_grad:
+                # remove as '.'-character is not allowed in buffers
+                s_name = name.replace('.', '')
+                self.m_name2s_name.update({name: s_name})
+                self.register_buffer(s_name, p.clone().detach().data)
+
+        self.collected_params = []
+
+    def reset_num_updates(self):
+        del self.num_updates
+        self.register_buffer('num_updates', torch.tensor(0, dtype=torch.int))
+
+    def forward(self, model):
+        decay = self.decay
+
+        if self.num_updates >= 0:
+            self.num_updates += 1
+            decay = min(self.decay, (1 + self.num_updates) / (10 + self.num_updates))
+
+        one_minus_decay = 1.0 - decay
+
+        with torch.no_grad():
+            m_param = dict(model.named_parameters())
+            shadow_params = dict(self.named_buffers())
+
+            for key in m_param:
+                if m_param[key].requires_grad:
+                    sname = self.m_name2s_name[key]
+                    shadow_params[sname] = shadow_params[sname].type_as(m_param[key])
+                    shadow_params[sname].sub_(one_minus_decay * (shadow_params[sname] - m_param[key]))
+                else:
+                    assert not key in self.m_name2s_name
+
+    def copy_to(self, model):
+        m_param = dict(model.named_parameters())
+        shadow_params = dict(self.named_buffers())
+        for key in m_param:
+            if m_param[key].requires_grad:
+                m_param[key].data.copy_(shadow_params[self.m_name2s_name[key]].data)
+            else:
+                assert not key in self.m_name2s_name
+
+    def store(self, parameters):
+        """
+        Save the current parameters for restoring later.
+        Args:
+          parameters: Iterable of `torch.nn.Parameter`; the parameters to be
+            temporarily stored.
+        """
+        self.collected_params = [param.clone() for param in parameters]
+
+    def restore(self, parameters):
+        """
+        Restore the parameters stored with the `store` method.
+        Useful to validate the model with EMA parameters without affecting the
+        original optimization process. Store the parameters before the
+        `copy_to` method. After validation (or model saving), use this to
+        restore the former parameters.
+        Args:
+          parameters: Iterable of `torch.nn.Parameter`; the parameters to be
+            updated with the stored parameters.
+        """
+        for c_param, param in zip(self.collected_params, parameters):
+            param.data.copy_(c_param.data)

backend/headless/fcbh/ldm/modules/encoders/noise_aug_modules.py

@@ -0,0 +1,35 @@
+from ..diffusionmodules.upscaling import ImageConcatWithNoiseAugmentation
+from ..diffusionmodules.openaimodel import Timestep
+import torch
+
+class CLIPEmbeddingNoiseAugmentation(ImageConcatWithNoiseAugmentation):
+    def __init__(self, *args, clip_stats_path=None, timestep_dim=256, **kwargs):
+        super().__init__(*args, **kwargs)
+        if clip_stats_path is None:
+            clip_mean, clip_std = torch.zeros(timestep_dim), torch.ones(timestep_dim)
+        else:
+            clip_mean, clip_std = torch.load(clip_stats_path, map_location="cpu")
+        self.register_buffer("data_mean", clip_mean[None, :], persistent=False)
+        self.register_buffer("data_std", clip_std[None, :], persistent=False)
+        self.time_embed = Timestep(timestep_dim)
+
+    def scale(self, x):
+        # re-normalize to centered mean and unit variance
+        x = (x - self.data_mean) * 1. / self.data_std
+        return x
+
+    def unscale(self, x):
+        # back to original data stats
+        x = (x * self.data_std) + self.data_mean
+        return x
+
+    def forward(self, x, noise_level=None):
+        if noise_level is None:
+            noise_level = torch.randint(0, self.max_noise_level, (x.shape[0],), device=x.device).long()
+        else:
+            assert isinstance(noise_level, torch.Tensor)
+        x = self.scale(x)
+        z = self.q_sample(x, noise_level)
+        z = self.unscale(z)
+        noise_level = self.time_embed(noise_level)
+        return z, noise_level

backend/headless/fcbh/ldm/modules/sub_quadratic_attention.py

@@ -0,0 +1,251 @@
+# original source:
+#   https://github.com/AminRezaei0x443/memory-efficient-attention/blob/1bc0d9e6ac5f82ea43a375135c4e1d3896ee1694/memory_efficient_attention/attention_torch.py
+# license:
+#   MIT
+# credit:
+#   Amin Rezaei (original author)
+#   Alex Birch (optimized algorithm for 3D tensors, at the expense of removing bias, masking and callbacks)
+# implementation of:
+#   Self-attention Does Not Need O(n2) Memory":
+#   https://arxiv.org/abs/2112.05682v2
+
+from functools import partial
+import torch
+from torch import Tensor
+from torch.utils.checkpoint import checkpoint
+import math
+
+try:
+	from typing import Optional, NamedTuple, List, Protocol
+except ImportError:
+	from typing import Optional, NamedTuple, List
+	from typing_extensions import Protocol
+
+from torch import Tensor
+from typing import List
+
+from fcbh import model_management
+
+def dynamic_slice(
+    x: Tensor,
+    starts: List[int],
+    sizes: List[int],
+) -> Tensor:
+    slicing = [slice(start, start + size) for start, size in zip(starts, sizes)]
+    return x[slicing]
+
+class AttnChunk(NamedTuple):
+    exp_values: Tensor
+    exp_weights_sum: Tensor
+    max_score: Tensor
+
+class SummarizeChunk(Protocol):
+    @staticmethod
+    def __call__(
+        query: Tensor,
+        key_t: Tensor,
+        value: Tensor,
+    ) -> AttnChunk: ...
+
+class ComputeQueryChunkAttn(Protocol):
+    @staticmethod
+    def __call__(
+        query: Tensor,
+        key_t: Tensor,
+        value: Tensor,
+    ) -> Tensor: ...
+
+def _summarize_chunk(
+    query: Tensor,
+    key_t: Tensor,
+    value: Tensor,
+    scale: float,
+    upcast_attention: bool,
+) -> AttnChunk:
+    if upcast_attention:
+        with torch.autocast(enabled=False, device_type = 'cuda'):
+            query = query.float()
+            key_t = key_t.float()
+            attn_weights = torch.baddbmm(
+                torch.empty(1, 1, 1, device=query.device, dtype=query.dtype),
+                query,
+                key_t,
+                alpha=scale,
+                beta=0,
+            )
+    else:
+        attn_weights = torch.baddbmm(
+            torch.empty(1, 1, 1, device=query.device, dtype=query.dtype),
+            query,
+            key_t,
+            alpha=scale,
+            beta=0,
+        )
+    max_score, _ = torch.max(attn_weights, -1, keepdim=True)
+    max_score = max_score.detach()
+    attn_weights -= max_score
+    torch.exp(attn_weights, out=attn_weights)
+    exp_weights = attn_weights.to(value.dtype)
+    exp_values = torch.bmm(exp_weights, value)
+    max_score = max_score.squeeze(-1)
+    return AttnChunk(exp_values, exp_weights.sum(dim=-1), max_score)
+
+def _query_chunk_attention(
+    query: Tensor,
+    key_t: Tensor,
+    value: Tensor,
+    summarize_chunk: SummarizeChunk,
+    kv_chunk_size: int,
+) -> Tensor:
+    batch_x_heads, k_channels_per_head, k_tokens = key_t.shape
+    _, _, v_channels_per_head = value.shape
+
+    def chunk_scanner(chunk_idx: int) -> AttnChunk:
+        key_chunk = dynamic_slice(
+            key_t,
+            (0, 0, chunk_idx),
+            (batch_x_heads, k_channels_per_head, kv_chunk_size)
+        )
+        value_chunk = dynamic_slice(
+            value,
+            (0, chunk_idx, 0),
+            (batch_x_heads, kv_chunk_size, v_channels_per_head)
+        )
+        return summarize_chunk(query, key_chunk, value_chunk)
+
+    chunks: List[AttnChunk] = [
+        chunk_scanner(chunk) for chunk in torch.arange(0, k_tokens, kv_chunk_size)
+    ]
+    acc_chunk = AttnChunk(*map(torch.stack, zip(*chunks)))
+    chunk_values, chunk_weights, chunk_max = acc_chunk
+
+    global_max, _ = torch.max(chunk_max, 0, keepdim=True)
+    max_diffs = torch.exp(chunk_max - global_max)
+    chunk_values *= torch.unsqueeze(max_diffs, -1)
+    chunk_weights *= max_diffs
+
+    all_values = chunk_values.sum(dim=0)
+    all_weights = torch.unsqueeze(chunk_weights, -1).sum(dim=0)
+    return all_values / all_weights
+
+# TODO: refactor CrossAttention#get_attention_scores to share code with this
+def _get_attention_scores_no_kv_chunking(
+    query: Tensor,
+    key_t: Tensor,
+    value: Tensor,
+    scale: float,
+    upcast_attention: bool,
+) -> Tensor:
+    if upcast_attention:
+        with torch.autocast(enabled=False, device_type = 'cuda'):
+            query = query.float()
+            key_t = key_t.float()
+            attn_scores = torch.baddbmm(
+                torch.empty(1, 1, 1, device=query.device, dtype=query.dtype),
+                query,
+                key_t,
+                alpha=scale,
+                beta=0,
+            )
+    else:
+        attn_scores = torch.baddbmm(
+            torch.empty(1, 1, 1, device=query.device, dtype=query.dtype),
+            query,
+            key_t,
+            alpha=scale,
+            beta=0,
+        )
+
+    try:
+        attn_probs = attn_scores.softmax(dim=-1)
+        del attn_scores
+    except model_management.OOM_EXCEPTION:
+        print("ran out of memory while running softmax in  _get_attention_scores_no_kv_chunking, trying slower in place softmax instead")
+        attn_scores -= attn_scores.max(dim=-1, keepdim=True).values
+        torch.exp(attn_scores, out=attn_scores)
+        summed = torch.sum(attn_scores, dim=-1, keepdim=True)
+        attn_scores /= summed
+        attn_probs = attn_scores
+
+    hidden_states_slice = torch.bmm(attn_probs.to(value.dtype), value)
+    return hidden_states_slice
+
+class ScannedChunk(NamedTuple):
+    chunk_idx: int
+    attn_chunk: AttnChunk
+
+def efficient_dot_product_attention(
+    query: Tensor,
+    key_t: Tensor,
+    value: Tensor,
+    query_chunk_size=1024,
+    kv_chunk_size: Optional[int] = None,
+    kv_chunk_size_min: Optional[int] = None,
+    use_checkpoint=True,
+    upcast_attention=False,
+):
+    """Computes efficient dot-product attention given query, transposed key, and value.
+      This is efficient version of attention presented in
+      https://arxiv.org/abs/2112.05682v2 which comes with O(sqrt(n)) memory requirements.
+      Args:
+        query: queries for calculating attention with shape of
+          `[batch * num_heads, tokens, channels_per_head]`.
+        key_t: keys for calculating attention with shape of
+          `[batch * num_heads, channels_per_head, tokens]`.
+        value: values to be used in attention with shape of
+          `[batch * num_heads, tokens, channels_per_head]`.
+        query_chunk_size: int: query chunks size
+        kv_chunk_size: Optional[int]: key/value chunks size. if None: defaults to sqrt(key_tokens)
+        kv_chunk_size_min: Optional[int]: key/value minimum chunk size. only considered when kv_chunk_size is None. changes `sqrt(key_tokens)` into `max(sqrt(key_tokens), kv_chunk_size_min)`, to ensure our chunk sizes don't get too small (smaller chunks = more chunks = less concurrent work done).
+        use_checkpoint: bool: whether to use checkpointing (recommended True for training, False for inference)
+      Returns:
+        Output of shape `[batch * num_heads, query_tokens, channels_per_head]`.
+      """
+    batch_x_heads, q_tokens, q_channels_per_head = query.shape
+    _, _, k_tokens = key_t.shape
+    scale = q_channels_per_head ** -0.5
+
+    kv_chunk_size = min(kv_chunk_size or int(math.sqrt(k_tokens)), k_tokens)
+    if kv_chunk_size_min is not None:
+        kv_chunk_size = max(kv_chunk_size, kv_chunk_size_min)
+
+    def get_query_chunk(chunk_idx: int) -> Tensor:
+        return dynamic_slice(
+            query,
+            (0, chunk_idx, 0),
+            (batch_x_heads, min(query_chunk_size, q_tokens), q_channels_per_head)
+        )
+    
+    summarize_chunk: SummarizeChunk = partial(_summarize_chunk, scale=scale, upcast_attention=upcast_attention)
+    summarize_chunk: SummarizeChunk = partial(checkpoint, summarize_chunk) if use_checkpoint else summarize_chunk
+    compute_query_chunk_attn: ComputeQueryChunkAttn = partial(
+        _get_attention_scores_no_kv_chunking,
+        scale=scale,
+        upcast_attention=upcast_attention
+    ) if k_tokens <= kv_chunk_size else (
+        # fast-path for when there's just 1 key-value chunk per query chunk (this is just sliced attention btw)
+        partial(
+            _query_chunk_attention,
+            kv_chunk_size=kv_chunk_size,
+            summarize_chunk=summarize_chunk,
+        )
+    )
+
+    if q_tokens <= query_chunk_size:
+        # fast-path for when there's just 1 query chunk
+        return compute_query_chunk_attn(
+            query=query,
+            key_t=key_t,
+            value=value,
+        )
+    
+    # TODO: maybe we should use torch.empty_like(query) to allocate storage in-advance,
+    # and pass slices to be mutated, instead of torch.cat()ing the returned slices
+    res = torch.cat([
+        compute_query_chunk_attn(
+            query=get_query_chunk(i * query_chunk_size),
+            key_t=key_t,
+            value=value,
+        ) for i in range(math.ceil(q_tokens / query_chunk_size))
+    ], dim=1)
+    return res

backend/headless/fcbh/ldm/util.py

@@ -0,0 +1,197 @@
+import importlib
+
+import torch
+from torch import optim
+import numpy as np
+
+from inspect import isfunction
+from PIL import Image, ImageDraw, ImageFont
+
+
+def log_txt_as_img(wh, xc, size=10):
+    # wh a tuple of (width, height)
+    # xc a list of captions to plot
+    b = len(xc)
+    txts = list()
+    for bi in range(b):
+        txt = Image.new("RGB", wh, color="white")
+        draw = ImageDraw.Draw(txt)
+        font = ImageFont.truetype('data/DejaVuSans.ttf', size=size)
+        nc = int(40 * (wh[0] / 256))
+        lines = "\n".join(xc[bi][start:start + nc] for start in range(0, len(xc[bi]), nc))
+
+        try:
+            draw.text((0, 0), lines, fill="black", font=font)
+        except UnicodeEncodeError:
+            print("Cant encode string for logging. Skipping.")
+
+        txt = np.array(txt).transpose(2, 0, 1) / 127.5 - 1.0
+        txts.append(txt)
+    txts = np.stack(txts)
+    txts = torch.tensor(txts)
+    return txts
+
+
+def ismap(x):
+    if not isinstance(x, torch.Tensor):
+        return False
+    return (len(x.shape) == 4) and (x.shape[1] > 3)
+
+
+def isimage(x):
+    if not isinstance(x,torch.Tensor):
+        return False
+    return (len(x.shape) == 4) and (x.shape[1] == 3 or x.shape[1] == 1)
+
+
+def exists(x):
+    return x is not None
+
+
+def default(val, d):
+    if exists(val):
+        return val
+    return d() if isfunction(d) else d
+
+
+def mean_flat(tensor):
+    """
+    https://github.com/openai/guided-diffusion/blob/27c20a8fab9cb472df5d6bdd6c8d11c8f430b924/guided_diffusion/nn.py#L86
+    Take the mean over all non-batch dimensions.
+    """
+    return tensor.mean(dim=list(range(1, len(tensor.shape))))
+
+
+def count_params(model, verbose=False):
+    total_params = sum(p.numel() for p in model.parameters())
+    if verbose:
+        print(f"{model.__class__.__name__} has {total_params*1.e-6:.2f} M params.")
+    return total_params
+
+
+def instantiate_from_config(config):
+    if not "target" in config:
+        if config == '__is_first_stage__':
+            return None
+        elif config == "__is_unconditional__":
+            return None
+        raise KeyError("Expected key `target` to instantiate.")
+    return get_obj_from_str(config["target"])(**config.get("params", dict()))
+
+
+def get_obj_from_str(string, reload=False):
+    module, cls = string.rsplit(".", 1)
+    if reload:
+        module_imp = importlib.import_module(module)
+        importlib.reload(module_imp)
+    return getattr(importlib.import_module(module, package=None), cls)
+
+
+class AdamWwithEMAandWings(optim.Optimizer):
+    # credit to https://gist.github.com/crowsonkb/65f7265353f403714fce3b2595e0b298
+    def __init__(self, params, lr=1.e-3, betas=(0.9, 0.999), eps=1.e-8,  # TODO: check hyperparameters before using
+                 weight_decay=1.e-2, amsgrad=False, ema_decay=0.9999,   # ema decay to match previous code
+                 ema_power=1., param_names=()):
+        """AdamW that saves EMA versions of the parameters."""
+        if not 0.0 <= lr:
+            raise ValueError("Invalid learning rate: {}".format(lr))
+        if not 0.0 <= eps:
+            raise ValueError("Invalid epsilon value: {}".format(eps))
+        if not 0.0 <= betas[0] < 1.0:
+            raise ValueError("Invalid beta parameter at index 0: {}".format(betas[0]))
+        if not 0.0 <= betas[1] < 1.0:
+            raise ValueError("Invalid beta parameter at index 1: {}".format(betas[1]))
+        if not 0.0 <= weight_decay:
+            raise ValueError("Invalid weight_decay value: {}".format(weight_decay))
+        if not 0.0 <= ema_decay <= 1.0:
+            raise ValueError("Invalid ema_decay value: {}".format(ema_decay))
+        defaults = dict(lr=lr, betas=betas, eps=eps,
+                        weight_decay=weight_decay, amsgrad=amsgrad, ema_decay=ema_decay,
+                        ema_power=ema_power, param_names=param_names)
+        super().__init__(params, defaults)
+
+    def __setstate__(self, state):
+        super().__setstate__(state)
+        for group in self.param_groups:
+            group.setdefault('amsgrad', False)
+
+    @torch.no_grad()
+    def step(self, closure=None):
+        """Performs a single optimization step.
+        Args:
+            closure (callable, optional): A closure that reevaluates the model
+                and returns the loss.
+        """
+        loss = None
+        if closure is not None:
+            with torch.enable_grad():
+                loss = closure()
+
+        for group in self.param_groups:
+            params_with_grad = []
+            grads = []
+            exp_avgs = []
+            exp_avg_sqs = []
+            ema_params_with_grad = []
+            state_sums = []
+            max_exp_avg_sqs = []
+            state_steps = []
+            amsgrad = group['amsgrad']
+            beta1, beta2 = group['betas']
+            ema_decay = group['ema_decay']
+            ema_power = group['ema_power']
+
+            for p in group['params']:
+                if p.grad is None:
+                    continue
+                params_with_grad.append(p)
+                if p.grad.is_sparse:
+                    raise RuntimeError('AdamW does not support sparse gradients')
+                grads.append(p.grad)
+
+                state = self.state[p]
+
+                # State initialization
+                if len(state) == 0:
+                    state['step'] = 0
+                    # Exponential moving average of gradient values
+                    state['exp_avg'] = torch.zeros_like(p, memory_format=torch.preserve_format)
+                    # Exponential moving average of squared gradient values
+                    state['exp_avg_sq'] = torch.zeros_like(p, memory_format=torch.preserve_format)
+                    if amsgrad:
+                        # Maintains max of all exp. moving avg. of sq. grad. values
+                        state['max_exp_avg_sq'] = torch.zeros_like(p, memory_format=torch.preserve_format)
+                    # Exponential moving average of parameter values
+                    state['param_exp_avg'] = p.detach().float().clone()
+
+                exp_avgs.append(state['exp_avg'])
+                exp_avg_sqs.append(state['exp_avg_sq'])
+                ema_params_with_grad.append(state['param_exp_avg'])
+
+                if amsgrad:
+                    max_exp_avg_sqs.append(state['max_exp_avg_sq'])
+
+                # update the steps for each param group update
+                state['step'] += 1
+                # record the step after step update
+                state_steps.append(state['step'])
+
+            optim._functional.adamw(params_with_grad,
+                    grads,
+                    exp_avgs,
+                    exp_avg_sqs,
+                    max_exp_avg_sqs,
+                    state_steps,
+                    amsgrad=amsgrad,
+                    beta1=beta1,
+                    beta2=beta2,
+                    lr=group['lr'],
+                    weight_decay=group['weight_decay'],
+                    eps=group['eps'],
+                    maximize=False)
+
+            cur_ema_decay = min(ema_decay, 1 - state['step'] ** -ema_power)
+            for param, ema_param in zip(params_with_grad, ema_params_with_grad):
+                ema_param.mul_(cur_ema_decay).add_(param.float(), alpha=1 - cur_ema_decay)
+
+        return loss

backend/headless/fcbh/lora.py

@@ -0,0 +1,213 @@
+import fcbh.utils
+
+LORA_CLIP_MAP = {
+    "mlp.fc1": "mlp_fc1",
+    "mlp.fc2": "mlp_fc2",
+    "self_attn.k_proj": "self_attn_k_proj",
+    "self_attn.q_proj": "self_attn_q_proj",
+    "self_attn.v_proj": "self_attn_v_proj",
+    "self_attn.out_proj": "self_attn_out_proj",
+}
+
+
+def load_lora(lora, to_load):
+    patch_dict = {}
+    loaded_keys = set()
+    for x in to_load:
+        alpha_name = "{}.alpha".format(x)
+        alpha = None
+        if alpha_name in lora.keys():
+            alpha = lora[alpha_name].item()
+            loaded_keys.add(alpha_name)
+
+        regular_lora = "{}.lora_up.weight".format(x)
+        diffusers_lora = "{}_lora.up.weight".format(x)
+        transformers_lora = "{}.lora_linear_layer.up.weight".format(x)
+        A_name = None
+
+        if regular_lora in lora.keys():
+            A_name = regular_lora
+            B_name = "{}.lora_down.weight".format(x)
+            mid_name = "{}.lora_mid.weight".format(x)
+        elif diffusers_lora in lora.keys():
+            A_name = diffusers_lora
+            B_name = "{}_lora.down.weight".format(x)
+            mid_name = None
+        elif transformers_lora in lora.keys():
+            A_name = transformers_lora
+            B_name ="{}.lora_linear_layer.down.weight".format(x)
+            mid_name = None
+
+        if A_name is not None:
+            mid = None
+            if mid_name is not None and mid_name in lora.keys():
+                mid = lora[mid_name]
+                loaded_keys.add(mid_name)
+            patch_dict[to_load[x]] = (lora[A_name], lora[B_name], alpha, mid)
+            loaded_keys.add(A_name)
+            loaded_keys.add(B_name)
+
+
+        ######## loha
+        hada_w1_a_name = "{}.hada_w1_a".format(x)
+        hada_w1_b_name = "{}.hada_w1_b".format(x)
+        hada_w2_a_name = "{}.hada_w2_a".format(x)
+        hada_w2_b_name = "{}.hada_w2_b".format(x)
+        hada_t1_name = "{}.hada_t1".format(x)
+        hada_t2_name = "{}.hada_t2".format(x)
+        if hada_w1_a_name in lora.keys():
+            hada_t1 = None
+            hada_t2 = None
+            if hada_t1_name in lora.keys():
+                hada_t1 = lora[hada_t1_name]
+                hada_t2 = lora[hada_t2_name]
+                loaded_keys.add(hada_t1_name)
+                loaded_keys.add(hada_t2_name)
+
+            patch_dict[to_load[x]] = (lora[hada_w1_a_name], lora[hada_w1_b_name], alpha, lora[hada_w2_a_name], lora[hada_w2_b_name], hada_t1, hada_t2)
+            loaded_keys.add(hada_w1_a_name)
+            loaded_keys.add(hada_w1_b_name)
+            loaded_keys.add(hada_w2_a_name)
+            loaded_keys.add(hada_w2_b_name)
+
+
+        ######## lokr
+        lokr_w1_name = "{}.lokr_w1".format(x)
+        lokr_w2_name = "{}.lokr_w2".format(x)
+        lokr_w1_a_name = "{}.lokr_w1_a".format(x)
+        lokr_w1_b_name = "{}.lokr_w1_b".format(x)
+        lokr_t2_name = "{}.lokr_t2".format(x)
+        lokr_w2_a_name = "{}.lokr_w2_a".format(x)
+        lokr_w2_b_name = "{}.lokr_w2_b".format(x)
+
+        lokr_w1 = None
+        if lokr_w1_name in lora.keys():
+            lokr_w1 = lora[lokr_w1_name]
+            loaded_keys.add(lokr_w1_name)
+
+        lokr_w2 = None
+        if lokr_w2_name in lora.keys():
+            lokr_w2 = lora[lokr_w2_name]
+            loaded_keys.add(lokr_w2_name)
+
+        lokr_w1_a = None
+        if lokr_w1_a_name in lora.keys():
+            lokr_w1_a = lora[lokr_w1_a_name]
+            loaded_keys.add(lokr_w1_a_name)
+
+        lokr_w1_b = None
+        if lokr_w1_b_name in lora.keys():
+            lokr_w1_b = lora[lokr_w1_b_name]
+            loaded_keys.add(lokr_w1_b_name)
+
+        lokr_w2_a = None
+        if lokr_w2_a_name in lora.keys():
+            lokr_w2_a = lora[lokr_w2_a_name]
+            loaded_keys.add(lokr_w2_a_name)
+
+        lokr_w2_b = None
+        if lokr_w2_b_name in lora.keys():
+            lokr_w2_b = lora[lokr_w2_b_name]
+            loaded_keys.add(lokr_w2_b_name)
+
+        lokr_t2 = None
+        if lokr_t2_name in lora.keys():
+            lokr_t2 = lora[lokr_t2_name]
+            loaded_keys.add(lokr_t2_name)
+
+        if (lokr_w1 is not None) or (lokr_w2 is not None) or (lokr_w1_a is not None) or (lokr_w2_a is not None):
+            patch_dict[to_load[x]] = (lokr_w1, lokr_w2, alpha, lokr_w1_a, lokr_w1_b, lokr_w2_a, lokr_w2_b, lokr_t2)
+
+
+        w_norm_name = "{}.w_norm".format(x)
+        b_norm_name = "{}.b_norm".format(x)
+        w_norm = lora.get(w_norm_name, None)
+        b_norm = lora.get(b_norm_name, None)
+
+        if w_norm is not None:
+            loaded_keys.add(w_norm_name)
+            patch_dict[to_load[x]] = (w_norm,)
+            if b_norm is not None:
+                loaded_keys.add(b_norm_name)
+                patch_dict["{}.bias".format(to_load[x][:-len(".weight")])] = (b_norm,)
+
+        diff_name = "{}.diff".format(x)
+        diff_weight = lora.get(diff_name, None)
+        if diff_weight is not None:
+            patch_dict[to_load[x]] = (diff_weight,)
+            loaded_keys.add(diff_name)
+
+        diff_bias_name = "{}.diff_b".format(x)
+        diff_bias = lora.get(diff_bias_name, None)
+        if diff_bias is not None:
+            patch_dict["{}.bias".format(to_load[x][:-len(".weight")])] = (diff_bias,)
+            loaded_keys.add(diff_bias_name)
+
+    for x in lora.keys():
+        if x not in loaded_keys:
+            print("lora key not loaded", x)
+    return patch_dict
+
+def model_lora_keys_clip(model, key_map={}):
+    sdk = model.state_dict().keys()
+
+    text_model_lora_key = "lora_te_text_model_encoder_layers_{}_{}"
+    clip_l_present = False
+    for b in range(32): #TODO: clean up
+        for c in LORA_CLIP_MAP:
+            k = "clip_h.transformer.text_model.encoder.layers.{}.{}.weight".format(b, c)
+            if k in sdk:
+                lora_key = text_model_lora_key.format(b, LORA_CLIP_MAP[c])
+                key_map[lora_key] = k
+                lora_key = "lora_te1_text_model_encoder_layers_{}_{}".format(b, LORA_CLIP_MAP[c])
+                key_map[lora_key] = k
+                lora_key = "text_encoder.text_model.encoder.layers.{}.{}".format(b, c) #diffusers lora
+                key_map[lora_key] = k
+
+            k = "clip_l.transformer.text_model.encoder.layers.{}.{}.weight".format(b, c)
+            if k in sdk:
+                lora_key = text_model_lora_key.format(b, LORA_CLIP_MAP[c])
+                key_map[lora_key] = k
+                lora_key = "lora_te1_text_model_encoder_layers_{}_{}".format(b, LORA_CLIP_MAP[c]) #SDXL base
+                key_map[lora_key] = k
+                clip_l_present = True
+                lora_key = "text_encoder.text_model.encoder.layers.{}.{}".format(b, c) #diffusers lora
+                key_map[lora_key] = k
+
+            k = "clip_g.transformer.text_model.encoder.layers.{}.{}.weight".format(b, c)
+            if k in sdk:
+                if clip_l_present:
+                    lora_key = "lora_te2_text_model_encoder_layers_{}_{}".format(b, LORA_CLIP_MAP[c]) #SDXL base
+                    key_map[lora_key] = k
+                    lora_key = "text_encoder_2.text_model.encoder.layers.{}.{}".format(b, c) #diffusers lora
+                    key_map[lora_key] = k
+                else:
+                    lora_key = "lora_te_text_model_encoder_layers_{}_{}".format(b, LORA_CLIP_MAP[c]) #TODO: test if this is correct for SDXL-Refiner
+                    key_map[lora_key] = k
+                    lora_key = "text_encoder.text_model.encoder.layers.{}.{}".format(b, c) #diffusers lora
+                    key_map[lora_key] = k
+
+    return key_map
+
+def model_lora_keys_unet(model, key_map={}):
+    sdk = model.state_dict().keys()
+
+    for k in sdk:
+        if k.startswith("diffusion_model.") and k.endswith(".weight"):
+            key_lora = k[len("diffusion_model."):-len(".weight")].replace(".", "_")
+            key_map["lora_unet_{}".format(key_lora)] = k
+
+    diffusers_keys = fcbh.utils.unet_to_diffusers(model.model_config.unet_config)
+    for k in diffusers_keys:
+        if k.endswith(".weight"):
+            unet_key = "diffusion_model.{}".format(diffusers_keys[k])
+            key_lora = k[:-len(".weight")].replace(".", "_")
+            key_map["lora_unet_{}".format(key_lora)] = unet_key
+
+            diffusers_lora_prefix = ["", "unet."]
+            for p in diffusers_lora_prefix:
+                diffusers_lora_key = "{}{}".format(p, k[:-len(".weight")].replace(".to_", ".processor.to_"))
+                if diffusers_lora_key.endswith(".to_out.0"):
+                    diffusers_lora_key = diffusers_lora_key[:-2]
+                key_map[diffusers_lora_key] = unet_key
+    return key_map

backend/headless/fcbh/model_base.py

@@ -0,0 +1,253 @@
+import torch
+from fcbh.ldm.modules.diffusionmodules.openaimodel import UNetModel
+from fcbh.ldm.modules.encoders.noise_aug_modules import CLIPEmbeddingNoiseAugmentation
+from fcbh.ldm.modules.diffusionmodules.openaimodel import Timestep
+import fcbh.model_management
+import fcbh.conds
+from enum import Enum
+from . import utils
+
+class ModelType(Enum):
+    EPS = 1
+    V_PREDICTION = 2
+
+
+from fcbh.model_sampling import EPS, V_PREDICTION, ModelSamplingDiscrete
+
+def model_sampling(model_config, model_type):
+    if model_type == ModelType.EPS:
+        c = EPS
+    elif model_type == ModelType.V_PREDICTION:
+        c = V_PREDICTION
+
+    s = ModelSamplingDiscrete
+
+    class ModelSampling(s, c):
+        pass
+
+    return ModelSampling(model_config)
+
+
+class BaseModel(torch.nn.Module):
+    def __init__(self, model_config, model_type=ModelType.EPS, device=None):
+        super().__init__()
+
+        unet_config = model_config.unet_config
+        self.latent_format = model_config.latent_format
+        self.model_config = model_config
+
+        if not unet_config.get("disable_unet_model_creation", False):
+            self.diffusion_model = UNetModel(**unet_config, device=device)
+        self.model_type = model_type
+        self.model_sampling = model_sampling(model_config, model_type)
+
+        self.adm_channels = unet_config.get("adm_in_channels", None)
+        if self.adm_channels is None:
+            self.adm_channels = 0
+        self.inpaint_model = False
+        print("model_type", model_type.name)
+        print("adm", self.adm_channels)
+
+    def apply_model(self, x, t, c_concat=None, c_crossattn=None, control=None, transformer_options={}, **kwargs):
+        sigma = t
+        xc = self.model_sampling.calculate_input(sigma, x)
+        if c_concat is not None:
+            xc = torch.cat([xc] + [c_concat], dim=1)
+
+        context = c_crossattn
+        dtype = self.get_dtype()
+        xc = xc.to(dtype)
+        t = self.model_sampling.timestep(t).float()
+        context = context.to(dtype)
+        extra_conds = {}
+        for o in kwargs:
+            extra = kwargs[o]
+            if hasattr(extra, "to"):
+                extra = extra.to(dtype)
+            extra_conds[o] = extra
+        model_output = self.diffusion_model(xc, t, context=context, control=control, transformer_options=transformer_options, **extra_conds).float()
+        return self.model_sampling.calculate_denoised(sigma, model_output, x)
+
+    def get_dtype(self):
+        return self.diffusion_model.dtype
+
+    def is_adm(self):
+        return self.adm_channels > 0
+
+    def encode_adm(self, **kwargs):
+        return None
+
+    def extra_conds(self, **kwargs):
+        out = {}
+        if self.inpaint_model:
+            concat_keys = ("mask", "masked_image")
+            cond_concat = []
+            denoise_mask = kwargs.get("denoise_mask", None)
+            latent_image = kwargs.get("latent_image", None)
+            noise = kwargs.get("noise", None)
+            device = kwargs["device"]
+
+            def blank_inpaint_image_like(latent_image):
+                blank_image = torch.ones_like(latent_image)
+                # these are the values for "zero" in pixel space translated to latent space
+                blank_image[:,0] *= 0.8223
+                blank_image[:,1] *= -0.6876
+                blank_image[:,2] *= 0.6364
+                blank_image[:,3] *= 0.1380
+                return blank_image
+
+            for ck in concat_keys:
+                if denoise_mask is not None:
+                    if ck == "mask":
+                        cond_concat.append(denoise_mask[:,:1].to(device))
+                    elif ck == "masked_image":
+                        cond_concat.append(latent_image.to(device)) #NOTE: the latent_image should be masked by the mask in pixel space
+                else:
+                    if ck == "mask":
+                        cond_concat.append(torch.ones_like(noise)[:,:1])
+                    elif ck == "masked_image":
+                        cond_concat.append(blank_inpaint_image_like(noise))
+            data = torch.cat(cond_concat, dim=1)
+            out['c_concat'] = fcbh.conds.CONDNoiseShape(data)
+        adm = self.encode_adm(**kwargs)
+        if adm is not None:
+            out['y'] = fcbh.conds.CONDRegular(adm)
+        return out
+
+    def load_model_weights(self, sd, unet_prefix=""):
+        to_load = {}
+        keys = list(sd.keys())
+        for k in keys:
+            if k.startswith(unet_prefix):
+                to_load[k[len(unet_prefix):]] = sd.pop(k)
+
+        m, u = self.diffusion_model.load_state_dict(to_load, strict=False)
+        if len(m) > 0:
+            print("unet missing:", m)
+
+        if len(u) > 0:
+            print("unet unexpected:", u)
+        del to_load
+        return self
+
+    def process_latent_in(self, latent):
+        return self.latent_format.process_in(latent)
+
+    def process_latent_out(self, latent):
+        return self.latent_format.process_out(latent)
+
+    def state_dict_for_saving(self, clip_state_dict, vae_state_dict):
+        clip_state_dict = self.model_config.process_clip_state_dict_for_saving(clip_state_dict)
+        unet_sd = self.diffusion_model.state_dict()
+        unet_state_dict = {}
+        for k in unet_sd:
+            unet_state_dict[k] = fcbh.model_management.resolve_lowvram_weight(unet_sd[k], self.diffusion_model, k)
+
+        unet_state_dict = self.model_config.process_unet_state_dict_for_saving(unet_state_dict)
+        vae_state_dict = self.model_config.process_vae_state_dict_for_saving(vae_state_dict)
+        if self.get_dtype() == torch.float16:
+            clip_state_dict = utils.convert_sd_to(clip_state_dict, torch.float16)
+            vae_state_dict = utils.convert_sd_to(vae_state_dict, torch.float16)
+
+        if self.model_type == ModelType.V_PREDICTION:
+            unet_state_dict["v_pred"] = torch.tensor([])
+
+        return {**unet_state_dict, **vae_state_dict, **clip_state_dict}
+
+    def set_inpaint(self):
+        self.inpaint_model = True
+
+def unclip_adm(unclip_conditioning, device, noise_augmentor, noise_augment_merge=0.0):
+    adm_inputs = []
+    weights = []
+    noise_aug = []
+    for unclip_cond in unclip_conditioning:
+        for adm_cond in unclip_cond["clip_vision_output"].image_embeds:
+            weight = unclip_cond["strength"]
+            noise_augment = unclip_cond["noise_augmentation"]
+            noise_level = round((noise_augmentor.max_noise_level - 1) * noise_augment)
+            c_adm, noise_level_emb = noise_augmentor(adm_cond.to(device), noise_level=torch.tensor([noise_level], device=device))
+            adm_out = torch.cat((c_adm, noise_level_emb), 1) * weight
+            weights.append(weight)
+            noise_aug.append(noise_augment)
+            adm_inputs.append(adm_out)
+
+    if len(noise_aug) > 1:
+        adm_out = torch.stack(adm_inputs).sum(0)
+        noise_augment = noise_augment_merge
+        noise_level = round((noise_augmentor.max_noise_level - 1) * noise_augment)
+        c_adm, noise_level_emb = noise_augmentor(adm_out[:, :noise_augmentor.time_embed.dim], noise_level=torch.tensor([noise_level], device=device))
+        adm_out = torch.cat((c_adm, noise_level_emb), 1)
+
+    return adm_out
+
+class SD21UNCLIP(BaseModel):
+    def __init__(self, model_config, noise_aug_config, model_type=ModelType.V_PREDICTION, device=None):
+        super().__init__(model_config, model_type, device=device)
+        self.noise_augmentor = CLIPEmbeddingNoiseAugmentation(**noise_aug_config)
+
+    def encode_adm(self, **kwargs):
+        unclip_conditioning = kwargs.get("unclip_conditioning", None)
+        device = kwargs["device"]
+        if unclip_conditioning is None:
+            return torch.zeros((1, self.adm_channels))
+        else:
+            return unclip_adm(unclip_conditioning, device, self.noise_augmentor, kwargs.get("unclip_noise_augment_merge", 0.05))
+
+def sdxl_pooled(args, noise_augmentor):
+    if "unclip_conditioning" in args:
+        return unclip_adm(args.get("unclip_conditioning", None), args["device"], noise_augmentor)[:,:1280]
+    else:
+        return args["pooled_output"]
+
+class SDXLRefiner(BaseModel):
+    def __init__(self, model_config, model_type=ModelType.EPS, device=None):
+        super().__init__(model_config, model_type, device=device)
+        self.embedder = Timestep(256)
+        self.noise_augmentor = CLIPEmbeddingNoiseAugmentation(**{"noise_schedule_config": {"timesteps": 1000, "beta_schedule": "squaredcos_cap_v2"}, "timestep_dim": 1280})
+
+    def encode_adm(self, **kwargs):
+        clip_pooled = sdxl_pooled(kwargs, self.noise_augmentor)
+        width = kwargs.get("width", 768)
+        height = kwargs.get("height", 768)
+        crop_w = kwargs.get("crop_w", 0)
+        crop_h = kwargs.get("crop_h", 0)
+
+        if kwargs.get("prompt_type", "") == "negative":
+            aesthetic_score = kwargs.get("aesthetic_score", 2.5)
+        else:
+            aesthetic_score = kwargs.get("aesthetic_score", 6)
+
+        out = []
+        out.append(self.embedder(torch.Tensor([height])))
+        out.append(self.embedder(torch.Tensor([width])))
+        out.append(self.embedder(torch.Tensor([crop_h])))
+        out.append(self.embedder(torch.Tensor([crop_w])))
+        out.append(self.embedder(torch.Tensor([aesthetic_score])))
+        flat = torch.flatten(torch.cat(out)).unsqueeze(dim=0).repeat(clip_pooled.shape[0], 1)
+        return torch.cat((clip_pooled.to(flat.device), flat), dim=1)
+
+class SDXL(BaseModel):
+    def __init__(self, model_config, model_type=ModelType.EPS, device=None):
+        super().__init__(model_config, model_type, device=device)
+        self.embedder = Timestep(256)
+        self.noise_augmentor = CLIPEmbeddingNoiseAugmentation(**{"noise_schedule_config": {"timesteps": 1000, "beta_schedule": "squaredcos_cap_v2"}, "timestep_dim": 1280})
+
+    def encode_adm(self, **kwargs):
+        clip_pooled = sdxl_pooled(kwargs, self.noise_augmentor)
+        width = kwargs.get("width", 768)
+        height = kwargs.get("height", 768)
+        crop_w = kwargs.get("crop_w", 0)
+        crop_h = kwargs.get("crop_h", 0)
+        target_width = kwargs.get("target_width", width)
+        target_height = kwargs.get("target_height", height)
+
+        out = []
+        out.append(self.embedder(torch.Tensor([height])))
+        out.append(self.embedder(torch.Tensor([width])))
+        out.append(self.embedder(torch.Tensor([crop_h])))
+        out.append(self.embedder(torch.Tensor([crop_w])))
+        out.append(self.embedder(torch.Tensor([target_height])))
+        out.append(self.embedder(torch.Tensor([target_width])))
+        flat = torch.flatten(torch.cat(out)).unsqueeze(dim=0).repeat(clip_pooled.shape[0], 1)
+        return torch.cat((clip_pooled.to(flat.device), flat), dim=1)

backend/headless/fcbh/model_detection.py

@@ -0,0 +1,270 @@
+import fcbh.supported_models
+import fcbh.supported_models_base
+
+def count_blocks(state_dict_keys, prefix_string):
+    count = 0
+    while True:
+        c = False
+        for k in state_dict_keys:
+            if k.startswith(prefix_string.format(count)):
+                c = True
+                break
+        if c == False:
+            break
+        count += 1
+    return count
+
+def calculate_transformer_depth(prefix, state_dict_keys, state_dict):
+    context_dim = None
+    use_linear_in_transformer = False
+
+    transformer_prefix = prefix + "1.transformer_blocks."
+    transformer_keys = sorted(list(filter(lambda a: a.startswith(transformer_prefix), state_dict_keys)))
+    if len(transformer_keys) > 0:
+        last_transformer_depth = count_blocks(state_dict_keys, transformer_prefix + '{}')
+        context_dim = state_dict['{}0.attn2.to_k.weight'.format(transformer_prefix)].shape[1]
+        use_linear_in_transformer = len(state_dict['{}1.proj_in.weight'.format(prefix)].shape) == 2
+        return last_transformer_depth, context_dim, use_linear_in_transformer
+    return None
+
+def detect_unet_config(state_dict, key_prefix, dtype):
+    state_dict_keys = list(state_dict.keys())
+
+    unet_config = {
+        "use_checkpoint": False,
+        "image_size": 32,
+        "out_channels": 4,
+        "use_spatial_transformer": True,
+        "legacy": False
+    }
+
+    y_input = '{}label_emb.0.0.weight'.format(key_prefix)
+    if y_input in state_dict_keys:
+        unet_config["num_classes"] = "sequential"
+        unet_config["adm_in_channels"] = state_dict[y_input].shape[1]
+    else:
+        unet_config["adm_in_channels"] = None
+
+    unet_config["dtype"] = dtype
+    model_channels = state_dict['{}input_blocks.0.0.weight'.format(key_prefix)].shape[0]
+    in_channels = state_dict['{}input_blocks.0.0.weight'.format(key_prefix)].shape[1]
+
+    num_res_blocks = []
+    channel_mult = []
+    attention_resolutions = []
+    transformer_depth = []
+    transformer_depth_output = []
+    context_dim = None
+    use_linear_in_transformer = False
+
+
+    current_res = 1
+    count = 0
+
+    last_res_blocks = 0
+    last_channel_mult = 0
+
+    input_block_count = count_blocks(state_dict_keys, '{}input_blocks'.format(key_prefix) + '.{}.')
+    for count in range(input_block_count):
+        prefix = '{}input_blocks.{}.'.format(key_prefix, count)
+        prefix_output = '{}output_blocks.{}.'.format(key_prefix, input_block_count - count - 1)
+
+        block_keys = sorted(list(filter(lambda a: a.startswith(prefix), state_dict_keys)))
+        if len(block_keys) == 0:
+            break
+
+        block_keys_output = sorted(list(filter(lambda a: a.startswith(prefix_output), state_dict_keys)))
+
+        if "{}0.op.weight".format(prefix) in block_keys: #new layer
+            num_res_blocks.append(last_res_blocks)
+            channel_mult.append(last_channel_mult)
+
+            current_res *= 2
+            last_res_blocks = 0
+            last_channel_mult = 0
+            out = calculate_transformer_depth(prefix_output, state_dict_keys, state_dict)
+            if out is not None:
+                transformer_depth_output.append(out[0])
+            else:
+                transformer_depth_output.append(0)
+        else:
+            res_block_prefix = "{}0.in_layers.0.weight".format(prefix)
+            if res_block_prefix in block_keys:
+                last_res_blocks += 1
+                last_channel_mult = state_dict["{}0.out_layers.3.weight".format(prefix)].shape[0] // model_channels
+
+                out = calculate_transformer_depth(prefix, state_dict_keys, state_dict)
+                if out is not None:
+                    transformer_depth.append(out[0])
+                    if context_dim is None:
+                        context_dim = out[1]
+                        use_linear_in_transformer = out[2]
+                else:
+                    transformer_depth.append(0)
+
+            res_block_prefix = "{}0.in_layers.0.weight".format(prefix_output)
+            if res_block_prefix in block_keys_output:
+                out = calculate_transformer_depth(prefix_output, state_dict_keys, state_dict)
+                if out is not None:
+                    transformer_depth_output.append(out[0])
+                else:
+                    transformer_depth_output.append(0)
+
+
+    num_res_blocks.append(last_res_blocks)
+    channel_mult.append(last_channel_mult)
+    if "{}middle_block.1.proj_in.weight".format(key_prefix) in state_dict_keys:
+        transformer_depth_middle = count_blocks(state_dict_keys, '{}middle_block.1.transformer_blocks.'.format(key_prefix) + '{}')
+    else:
+        transformer_depth_middle = -1
+
+    unet_config["in_channels"] = in_channels
+    unet_config["model_channels"] = model_channels
+    unet_config["num_res_blocks"] = num_res_blocks
+    unet_config["transformer_depth"] = transformer_depth
+    unet_config["transformer_depth_output"] = transformer_depth_output
+    unet_config["channel_mult"] = channel_mult
+    unet_config["transformer_depth_middle"] = transformer_depth_middle
+    unet_config['use_linear_in_transformer'] = use_linear_in_transformer
+    unet_config["context_dim"] = context_dim
+    return unet_config
+
+def model_config_from_unet_config(unet_config):
+    for model_config in fcbh.supported_models.models:
+        if model_config.matches(unet_config):
+            return model_config(unet_config)
+
+    print("no match", unet_config)
+    return None
+
+def model_config_from_unet(state_dict, unet_key_prefix, dtype, use_base_if_no_match=False):
+    unet_config = detect_unet_config(state_dict, unet_key_prefix, dtype)
+    model_config = model_config_from_unet_config(unet_config)
+    if model_config is None and use_base_if_no_match:
+        return fcbh.supported_models_base.BASE(unet_config)
+    else:
+        return model_config
+
+def convert_config(unet_config):
+    new_config = unet_config.copy()
+    num_res_blocks = new_config.get("num_res_blocks", None)
+    channel_mult = new_config.get("channel_mult", None)
+
+    if isinstance(num_res_blocks, int):
+        num_res_blocks = len(channel_mult) * [num_res_blocks]
+
+    if "attention_resolutions" in new_config:
+        attention_resolutions = new_config.pop("attention_resolutions")
+        transformer_depth = new_config.get("transformer_depth", None)
+        transformer_depth_middle = new_config.get("transformer_depth_middle", None)
+
+        if isinstance(transformer_depth, int):
+            transformer_depth = len(channel_mult) * [transformer_depth]
+        if transformer_depth_middle is None:
+            transformer_depth_middle =  transformer_depth[-1]
+        t_in = []
+        t_out = []
+        s = 1
+        for i in range(len(num_res_blocks)):
+            res = num_res_blocks[i]
+            d = 0
+            if s in attention_resolutions:
+                d = transformer_depth[i]
+
+            t_in += [d] * res
+            t_out += [d] * (res + 1)
+            s *= 2
+        transformer_depth = t_in
+        transformer_depth_output = t_out
+        new_config["transformer_depth"] = t_in
+        new_config["transformer_depth_output"] = t_out
+        new_config["transformer_depth_middle"] = transformer_depth_middle
+
+    new_config["num_res_blocks"] = num_res_blocks
+    return new_config
+
+
+def unet_config_from_diffusers_unet(state_dict, dtype):
+    match = {}
+    attention_resolutions = []
+
+    attn_res = 1
+    for i in range(5):
+        k = "down_blocks.{}.attentions.1.transformer_blocks.0.attn2.to_k.weight".format(i)
+        if k in state_dict:
+            match["context_dim"] = state_dict[k].shape[1]
+            attention_resolutions.append(attn_res)
+        attn_res *= 2
+
+    match["attention_resolutions"] = attention_resolutions
+
+    match["model_channels"] = state_dict["conv_in.weight"].shape[0]
+    match["in_channels"] = state_dict["conv_in.weight"].shape[1]
+    match["adm_in_channels"] = None
+    if "class_embedding.linear_1.weight" in state_dict:
+        match["adm_in_channels"] = state_dict["class_embedding.linear_1.weight"].shape[1]
+    elif "add_embedding.linear_1.weight" in state_dict:
+        match["adm_in_channels"] = state_dict["add_embedding.linear_1.weight"].shape[1]
+
+    SDXL = {'use_checkpoint': False, 'image_size': 32, 'out_channels': 4, 'use_spatial_transformer': True, 'legacy': False,
+            'num_classes': 'sequential', 'adm_in_channels': 2816, 'dtype': dtype, 'in_channels': 4, 'model_channels': 320,
+            'num_res_blocks': 2, 'attention_resolutions': [2, 4], 'transformer_depth': [0, 2, 10], 'channel_mult': [1, 2, 4],
+            'transformer_depth_middle': 10, 'use_linear_in_transformer': True, 'context_dim': 2048, "num_head_channels": 64}
+
+    SDXL_refiner = {'use_checkpoint': False, 'image_size': 32, 'out_channels': 4, 'use_spatial_transformer': True, 'legacy': False,
+                    'num_classes': 'sequential', 'adm_in_channels': 2560, 'dtype': dtype, 'in_channels': 4, 'model_channels': 384,
+                    'num_res_blocks': 2, 'attention_resolutions': [2, 4], 'transformer_depth': [0, 4, 4, 0], 'channel_mult': [1, 2, 4, 4],
+                    'transformer_depth_middle': 4, 'use_linear_in_transformer': True, 'context_dim': 1280, "num_head_channels": 64}
+
+    SD21 = {'use_checkpoint': False, 'image_size': 32, 'out_channels': 4, 'use_spatial_transformer': True, 'legacy': False,
+            'adm_in_channels': None, 'dtype': dtype, 'in_channels': 4, 'model_channels': 320, 'num_res_blocks': 2,
+            'attention_resolutions': [1, 2, 4], 'transformer_depth': [1, 1, 1, 0], 'channel_mult': [1, 2, 4, 4],
+            'transformer_depth_middle': 1, 'use_linear_in_transformer': True, 'context_dim': 1024, "num_head_channels": 64}
+
+    SD21_uncliph = {'use_checkpoint': False, 'image_size': 32, 'out_channels': 4, 'use_spatial_transformer': True, 'legacy': False,
+                    'num_classes': 'sequential', 'adm_in_channels': 2048, 'dtype': dtype, 'in_channels': 4, 'model_channels': 320,
+                    'num_res_blocks': 2, 'attention_resolutions': [1, 2, 4], 'transformer_depth': [1, 1, 1, 0], 'channel_mult': [1, 2, 4, 4],
+                    'transformer_depth_middle': 1, 'use_linear_in_transformer': True, 'context_dim': 1024, "num_head_channels": 64}
+
+    SD21_unclipl = {'use_checkpoint': False, 'image_size': 32, 'out_channels': 4, 'use_spatial_transformer': True, 'legacy': False,
+                    'num_classes': 'sequential', 'adm_in_channels': 1536, 'dtype': dtype, 'in_channels': 4, 'model_channels': 320,
+                    'num_res_blocks': 2, 'attention_resolutions': [1, 2, 4], 'transformer_depth': [1, 1, 1, 0], 'channel_mult': [1, 2, 4, 4],
+                    'transformer_depth_middle': 1, 'use_linear_in_transformer': True, 'context_dim': 1024}
+
+    SD15 = {'use_checkpoint': False, 'image_size': 32, 'out_channels': 4, 'use_spatial_transformer': True, 'legacy': False,
+            'adm_in_channels': None, 'dtype': dtype, 'in_channels': 4, 'model_channels': 320, 'num_res_blocks': 2,
+            'attention_resolutions': [1, 2, 4], 'transformer_depth': [1, 1, 1, 0], 'channel_mult': [1, 2, 4, 4],
+            'transformer_depth_middle': 1, 'use_linear_in_transformer': False, 'context_dim': 768, "num_heads": 8}
+
+    SDXL_mid_cnet = {'use_checkpoint': False, 'image_size': 32, 'out_channels': 4, 'use_spatial_transformer': True, 'legacy': False,
+            'num_classes': 'sequential', 'adm_in_channels': 2816, 'dtype': dtype, 'in_channels': 4, 'model_channels': 320,
+            'num_res_blocks': 2, 'attention_resolutions': [4], 'transformer_depth': [0, 0, 1], 'channel_mult': [1, 2, 4],
+            'transformer_depth_middle': 1, 'use_linear_in_transformer': True, 'context_dim': 2048, "num_head_channels": 64}
+
+    SDXL_small_cnet = {'use_checkpoint': False, 'image_size': 32, 'out_channels': 4, 'use_spatial_transformer': True, 'legacy': False,
+            'num_classes': 'sequential', 'adm_in_channels': 2816, 'dtype': dtype, 'in_channels': 4, 'model_channels': 320,
+            'num_res_blocks': 2, 'attention_resolutions': [], 'transformer_depth': [0, 0, 0], 'channel_mult': [1, 2, 4],
+            'transformer_depth_middle': 0, 'use_linear_in_transformer': True, "num_head_channels": 64, 'context_dim': 1}
+
+    SDXL_diffusers_inpaint = {'use_checkpoint': False, 'image_size': 32, 'out_channels': 4, 'use_spatial_transformer': True, 'legacy': False,
+            'num_classes': 'sequential', 'adm_in_channels': 2816, 'dtype': dtype, 'in_channels': 9, 'model_channels': 320,
+            'num_res_blocks': 2, 'attention_resolutions': [2, 4], 'transformer_depth': [0, 2, 10], 'channel_mult': [1, 2, 4],
+            'transformer_depth_middle': 10, 'use_linear_in_transformer': True, 'context_dim': 2048, "num_head_channels": 64}
+
+    supported_models = [SDXL, SDXL_refiner, SD21, SD15, SD21_uncliph, SD21_unclipl, SDXL_mid_cnet, SDXL_small_cnet, SDXL_diffusers_inpaint]
+
+    for unet_config in supported_models:
+        matches = True
+        for k in match:
+            if match[k] != unet_config[k]:
+                matches = False
+                break
+        if matches:
+            return convert_config(unet_config)
+    return None
+
+def model_config_from_diffusers_unet(state_dict, dtype):
+    unet_config = unet_config_from_diffusers_unet(state_dict, dtype)
+    if unet_config is not None:
+        return model_config_from_unet_config(unet_config)
+    return None

backend/headless/fcbh/model_management.py

@@ -0,0 +1,726 @@
+import psutil
+from enum import Enum
+from fcbh.cli_args import args
+import fcbh.utils
+import torch
+import sys
+
+class VRAMState(Enum):
+    DISABLED = 0    #No vram present: no need to move models to vram
+    NO_VRAM = 1     #Very low vram: enable all the options to save vram
+    LOW_VRAM = 2
+    NORMAL_VRAM = 3
+    HIGH_VRAM = 4
+    SHARED = 5      #No dedicated vram: memory shared between CPU and GPU but models still need to be moved between both.
+
+class CPUState(Enum):
+    GPU = 0
+    CPU = 1
+    MPS = 2
+
+# Determine VRAM State
+vram_state = VRAMState.NORMAL_VRAM
+set_vram_to = VRAMState.NORMAL_VRAM
+cpu_state = CPUState.GPU
+
+total_vram = 0
+
+lowvram_available = True
+xpu_available = False
+
+directml_enabled = False
+if args.directml is not None:
+    import torch_directml
+    directml_enabled = True
+    device_index = args.directml
+    if device_index < 0:
+        directml_device = torch_directml.device()
+    else:
+        directml_device = torch_directml.device(device_index)
+    print("Using directml with device:", torch_directml.device_name(device_index))
+    # torch_directml.disable_tiled_resources(True)
+    lowvram_available = False #TODO: need to find a way to get free memory in directml before this can be enabled by default.
+
+try:
+    import intel_extension_for_pytorch as ipex
+    if torch.xpu.is_available():
+        xpu_available = True
+except:
+    pass
+
+try:
+    if torch.backends.mps.is_available():
+        cpu_state = CPUState.MPS
+        import torch.mps
+except:
+    pass
+
+if args.cpu:
+    cpu_state = CPUState.CPU
+
+def is_intel_xpu():
+    global cpu_state
+    global xpu_available
+    if cpu_state == CPUState.GPU:
+        if xpu_available:
+            return True
+    return False
+
+def get_torch_device():
+    global directml_enabled
+    global cpu_state
+    if directml_enabled:
+        global directml_device
+        return directml_device
+    if cpu_state == CPUState.MPS:
+        return torch.device("mps")
+    if cpu_state == CPUState.CPU:
+        return torch.device("cpu")
+    else:
+        if is_intel_xpu():
+            return torch.device("xpu")
+        else:
+            return torch.device(torch.cuda.current_device())
+
+def get_total_memory(dev=None, torch_total_too=False):
+    global directml_enabled
+    if dev is None:
+        dev = get_torch_device()
+
+    if hasattr(dev, 'type') and (dev.type == 'cpu' or dev.type == 'mps'):
+        mem_total = psutil.virtual_memory().total
+        mem_total_torch = mem_total
+    else:
+        if directml_enabled:
+            mem_total = 1024 * 1024 * 1024 #TODO
+            mem_total_torch = mem_total
+        elif is_intel_xpu():
+            stats = torch.xpu.memory_stats(dev)
+            mem_reserved = stats['reserved_bytes.all.current']
+            mem_total = torch.xpu.get_device_properties(dev).total_memory
+            mem_total_torch = mem_reserved
+        else:
+            stats = torch.cuda.memory_stats(dev)
+            mem_reserved = stats['reserved_bytes.all.current']
+            _, mem_total_cuda = torch.cuda.mem_get_info(dev)
+            mem_total_torch = mem_reserved
+            mem_total = mem_total_cuda
+
+    if torch_total_too:
+        return (mem_total, mem_total_torch)
+    else:
+        return mem_total
+
+total_vram = get_total_memory(get_torch_device()) / (1024 * 1024)
+total_ram = psutil.virtual_memory().total / (1024 * 1024)
+print("Total VRAM {:0.0f} MB, total RAM {:0.0f} MB".format(total_vram, total_ram))
+if not args.normalvram and not args.cpu:
+    if lowvram_available and total_vram <= 4096:
+        print("Trying to enable lowvram mode because your GPU seems to have 4GB or less. If you don't want this use: --normalvram")
+        set_vram_to = VRAMState.LOW_VRAM
+
+try:
+    OOM_EXCEPTION = torch.cuda.OutOfMemoryError
+except:
+    OOM_EXCEPTION = Exception
+
+XFORMERS_VERSION = ""
+XFORMERS_ENABLED_VAE = True
+if args.disable_xformers:
+    XFORMERS_IS_AVAILABLE = False
+else:
+    try:
+        import xformers
+        import xformers.ops
+        XFORMERS_IS_AVAILABLE = True
+        try:
+            XFORMERS_VERSION = xformers.version.__version__
+            print("xformers version:", XFORMERS_VERSION)
+            if XFORMERS_VERSION.startswith("0.0.18"):
+                print()
+                print("WARNING: This version of xformers has a major bug where you will get black images when generating high resolution images.")
+                print("Please downgrade or upgrade xformers to a different version.")
+                print()
+                XFORMERS_ENABLED_VAE = False
+        except:
+            pass
+    except:
+        XFORMERS_IS_AVAILABLE = False
+
+def is_nvidia():
+    global cpu_state
+    if cpu_state == CPUState.GPU:
+        if torch.version.cuda:
+            return True
+    return False
+
+ENABLE_PYTORCH_ATTENTION = False
+if args.use_pytorch_cross_attention:
+    ENABLE_PYTORCH_ATTENTION = True
+    XFORMERS_IS_AVAILABLE = False
+
+VAE_DTYPE = torch.float32
+
+try:
+    if is_nvidia():
+        torch_version = torch.version.__version__
+        if int(torch_version[0]) >= 2:
+            if ENABLE_PYTORCH_ATTENTION == False and args.use_split_cross_attention == False and args.use_quad_cross_attention == False:
+                ENABLE_PYTORCH_ATTENTION = True
+            if torch.cuda.is_bf16_supported():
+                VAE_DTYPE = torch.bfloat16
+    if is_intel_xpu():
+        if args.use_split_cross_attention == False and args.use_quad_cross_attention == False:
+            ENABLE_PYTORCH_ATTENTION = True
+except:
+    pass
+
+if is_intel_xpu():
+    VAE_DTYPE = torch.bfloat16
+
+if args.fp16_vae:
+    VAE_DTYPE = torch.float16
+elif args.bf16_vae:
+    VAE_DTYPE = torch.bfloat16
+elif args.fp32_vae:
+    VAE_DTYPE = torch.float32
+
+
+if ENABLE_PYTORCH_ATTENTION:
+    torch.backends.cuda.enable_math_sdp(True)
+    torch.backends.cuda.enable_flash_sdp(True)
+    torch.backends.cuda.enable_mem_efficient_sdp(True)
+
+if args.lowvram:
+    set_vram_to = VRAMState.LOW_VRAM
+    lowvram_available = True
+elif args.novram:
+    set_vram_to = VRAMState.NO_VRAM
+elif args.highvram or args.gpu_only:
+    vram_state = VRAMState.HIGH_VRAM
+
+FORCE_FP32 = False
+FORCE_FP16 = False
+if args.force_fp32:
+    print("Forcing FP32, if this improves things please report it.")
+    FORCE_FP32 = True
+
+if args.force_fp16:
+    print("Forcing FP16.")
+    FORCE_FP16 = True
+
+if lowvram_available:
+    try:
+        import accelerate
+        if set_vram_to in (VRAMState.LOW_VRAM, VRAMState.NO_VRAM):
+            vram_state = set_vram_to
+    except Exception as e:
+        import traceback
+        print(traceback.format_exc())
+        print("ERROR: LOW VRAM MODE NEEDS accelerate.")
+        lowvram_available = False
+
+
+if cpu_state != CPUState.GPU:
+    vram_state = VRAMState.DISABLED
+
+if cpu_state == CPUState.MPS:
+    vram_state = VRAMState.SHARED
+
+print(f"Set vram state to: {vram_state.name}")
+
+DISABLE_SMART_MEMORY = args.disable_smart_memory
+
+if DISABLE_SMART_MEMORY:
+    print("Disabling smart memory management")
+
+def get_torch_device_name(device):
+    if hasattr(device, 'type'):
+        if device.type == "cuda":
+            try:
+                allocator_backend = torch.cuda.get_allocator_backend()
+            except:
+                allocator_backend = ""
+            return "{} {} : {}".format(device, torch.cuda.get_device_name(device), allocator_backend)
+        else:
+            return "{}".format(device.type)
+    elif is_intel_xpu():
+        return "{} {}".format(device, torch.xpu.get_device_name(device))
+    else:
+        return "CUDA {}: {}".format(device, torch.cuda.get_device_name(device))
+
+try:
+    print("Device:", get_torch_device_name(get_torch_device()))
+except:
+    print("Could not pick default device.")
+
+print("VAE dtype:", VAE_DTYPE)
+
+current_loaded_models = []
+
+class LoadedModel:
+    def __init__(self, model):
+        self.model = model
+        self.model_accelerated = False
+        self.device = model.load_device
+
+    def model_memory(self):
+        return self.model.model_size()
+
+    def model_memory_required(self, device):
+        if device == self.model.current_device:
+            return 0
+        else:
+            return self.model_memory()
+
+    def model_load(self, lowvram_model_memory=0):
+        patch_model_to = None
+        if lowvram_model_memory == 0:
+            patch_model_to = self.device
+
+        self.model.model_patches_to(self.device)
+        self.model.model_patches_to(self.model.model_dtype())
+
+        try:
+            self.real_model = self.model.patch_model(device_to=patch_model_to) #TODO: do something with loras and offloading to CPU
+        except Exception as e:
+            self.model.unpatch_model(self.model.offload_device)
+            self.model_unload()
+            raise e
+
+        if lowvram_model_memory > 0:
+            print("loading in lowvram mode", lowvram_model_memory/(1024 * 1024))
+            device_map = accelerate.infer_auto_device_map(self.real_model, max_memory={0: "{}MiB".format(lowvram_model_memory // (1024 * 1024)), "cpu": "16GiB"})
+            accelerate.dispatch_model(self.real_model, device_map=device_map, main_device=self.device)
+            self.model_accelerated = True
+
+        if is_intel_xpu() and not args.disable_ipex_optimize:
+            self.real_model = torch.xpu.optimize(self.real_model.eval(), inplace=True, auto_kernel_selection=True, graph_mode=True)
+
+        return self.real_model
+
+    def model_unload(self):
+        if self.model_accelerated:
+            accelerate.hooks.remove_hook_from_submodules(self.real_model)
+            self.model_accelerated = False
+
+        self.model.unpatch_model(self.model.offload_device)
+        self.model.model_patches_to(self.model.offload_device)
+
+    def __eq__(self, other):
+        return self.model is other.model
+
+def minimum_inference_memory():
+    return (1024 * 1024 * 1024)
+
+def unload_model_clones(model):
+    to_unload = []
+    for i in range(len(current_loaded_models)):
+        if model.is_clone(current_loaded_models[i].model):
+            to_unload = [i] + to_unload
+
+    for i in to_unload:
+        print("unload clone", i)
+        current_loaded_models.pop(i).model_unload()
+
+def free_memory(memory_required, device, keep_loaded=[]):
+    unloaded_model = False
+    for i in range(len(current_loaded_models) -1, -1, -1):
+        if not DISABLE_SMART_MEMORY:
+            if get_free_memory(device) > memory_required:
+                break
+        shift_model = current_loaded_models[i]
+        if shift_model.device == device:
+            if shift_model not in keep_loaded:
+                m = current_loaded_models.pop(i)
+                m.model_unload()
+                del m
+                unloaded_model = True
+
+    if unloaded_model:
+        soft_empty_cache()
+    else:
+        if vram_state != VRAMState.HIGH_VRAM:
+            mem_free_total, mem_free_torch = get_free_memory(device, torch_free_too=True)
+            if mem_free_torch > mem_free_total * 0.25:
+                soft_empty_cache()
+
+def load_models_gpu(models, memory_required=0):
+    global vram_state
+
+    inference_memory = minimum_inference_memory()
+    extra_mem = max(inference_memory, memory_required)
+
+    models_to_load = []
+    models_already_loaded = []
+    for x in models:
+        loaded_model = LoadedModel(x)
+
+        if loaded_model in current_loaded_models:
+            index = current_loaded_models.index(loaded_model)
+            current_loaded_models.insert(0, current_loaded_models.pop(index))
+            models_already_loaded.append(loaded_model)
+        else:
+            if hasattr(x, "model"):
+                print(f"Requested to load {x.model.__class__.__name__}")
+            models_to_load.append(loaded_model)
+
+    if len(models_to_load) == 0:
+        devs = set(map(lambda a: a.device, models_already_loaded))
+        for d in devs:
+            if d != torch.device("cpu"):
+                free_memory(extra_mem, d, models_already_loaded)
+        return
+
+    print(f"Loading {len(models_to_load)} new model{'s' if len(models_to_load) > 1 else ''}")
+
+    total_memory_required = {}
+    for loaded_model in models_to_load:
+        unload_model_clones(loaded_model.model)
+        total_memory_required[loaded_model.device] = total_memory_required.get(loaded_model.device, 0) + loaded_model.model_memory_required(loaded_model.device)
+
+    for device in total_memory_required:
+        if device != torch.device("cpu"):
+            free_memory(total_memory_required[device] * 1.3 + extra_mem, device, models_already_loaded)
+
+    for loaded_model in models_to_load:
+        model = loaded_model.model
+        torch_dev = model.load_device
+        if is_device_cpu(torch_dev):
+            vram_set_state = VRAMState.DISABLED
+        else:
+            vram_set_state = vram_state
+        lowvram_model_memory = 0
+        if lowvram_available and (vram_set_state == VRAMState.LOW_VRAM or vram_set_state == VRAMState.NORMAL_VRAM):
+            model_size = loaded_model.model_memory_required(torch_dev)
+            current_free_mem = get_free_memory(torch_dev)
+            lowvram_model_memory = int(max(256 * (1024 * 1024), (current_free_mem - 1024 * (1024 * 1024)) / 1.3 ))
+            if model_size > (current_free_mem - inference_memory): #only switch to lowvram if really necessary
+                vram_set_state = VRAMState.LOW_VRAM
+            else:
+                lowvram_model_memory = 0
+
+        if vram_set_state == VRAMState.NO_VRAM:
+            lowvram_model_memory = 256 * 1024 * 1024
+
+        cur_loaded_model = loaded_model.model_load(lowvram_model_memory)
+        current_loaded_models.insert(0, loaded_model)
+    return
+
+
+def load_model_gpu(model):
+    return load_models_gpu([model])
+
+def cleanup_models():
+    to_delete = []
+    for i in range(len(current_loaded_models)):
+        if sys.getrefcount(current_loaded_models[i].model) <= 2:
+            to_delete = [i] + to_delete
+
+    for i in to_delete:
+        x = current_loaded_models.pop(i)
+        x.model_unload()
+        del x
+
+def dtype_size(dtype):
+    dtype_size = 4
+    if dtype == torch.float16 or dtype == torch.bfloat16:
+        dtype_size = 2
+    return dtype_size
+
+def unet_offload_device():
+    if vram_state == VRAMState.HIGH_VRAM:
+        return get_torch_device()
+    else:
+        return torch.device("cpu")
+
+def unet_inital_load_device(parameters, dtype):
+    torch_dev = get_torch_device()
+    if vram_state == VRAMState.HIGH_VRAM:
+        return torch_dev
+
+    cpu_dev = torch.device("cpu")
+    if DISABLE_SMART_MEMORY:
+        return cpu_dev
+
+    model_size = dtype_size(dtype) * parameters
+
+    mem_dev = get_free_memory(torch_dev)
+    mem_cpu = get_free_memory(cpu_dev)
+    if mem_dev > mem_cpu and model_size < mem_dev:
+        return torch_dev
+    else:
+        return cpu_dev
+
+def unet_dtype(device=None, model_params=0):
+    if args.bf16_unet:
+        return torch.bfloat16
+    if should_use_fp16(device=device, model_params=model_params):
+        return torch.float16
+    return torch.float32
+
+def text_encoder_offload_device():
+    if args.gpu_only:
+        return get_torch_device()
+    else:
+        return torch.device("cpu")
+
+def text_encoder_device():
+    if args.gpu_only:
+        return get_torch_device()
+    elif vram_state == VRAMState.HIGH_VRAM or vram_state == VRAMState.NORMAL_VRAM:
+        if is_intel_xpu():
+            return torch.device("cpu")
+        if should_use_fp16(prioritize_performance=False):
+            return get_torch_device()
+        else:
+            return torch.device("cpu")
+    else:
+        return torch.device("cpu")
+
+def vae_device():
+    return get_torch_device()
+
+def vae_offload_device():
+    if args.gpu_only:
+        return get_torch_device()
+    else:
+        return torch.device("cpu")
+
+def vae_dtype():
+    global VAE_DTYPE
+    return VAE_DTYPE
+
+def get_autocast_device(dev):
+    if hasattr(dev, 'type'):
+        return dev.type
+    return "cuda"
+
+def cast_to_device(tensor, device, dtype, copy=False):
+    device_supports_cast = False
+    if tensor.dtype == torch.float32 or tensor.dtype == torch.float16:
+        device_supports_cast = True
+    elif tensor.dtype == torch.bfloat16:
+        if hasattr(device, 'type') and device.type.startswith("cuda"):
+            device_supports_cast = True
+        elif is_intel_xpu():
+            device_supports_cast = True
+
+    if device_supports_cast:
+        if copy:
+            if tensor.device == device:
+                return tensor.to(dtype, copy=copy)
+            return tensor.to(device, copy=copy).to(dtype)
+        else:
+            return tensor.to(device).to(dtype)
+    else:
+        return tensor.to(dtype).to(device, copy=copy)
+
+def xformers_enabled():
+    global directml_enabled
+    global cpu_state
+    if cpu_state != CPUState.GPU:
+        return False
+    if is_intel_xpu():
+        return False
+    if directml_enabled:
+        return False
+    return XFORMERS_IS_AVAILABLE
+
+
+def xformers_enabled_vae():
+    enabled = xformers_enabled()
+    if not enabled:
+        return False
+
+    return XFORMERS_ENABLED_VAE
+
+def pytorch_attention_enabled():
+    global ENABLE_PYTORCH_ATTENTION
+    return ENABLE_PYTORCH_ATTENTION
+
+def pytorch_attention_flash_attention():
+    global ENABLE_PYTORCH_ATTENTION
+    if ENABLE_PYTORCH_ATTENTION:
+        #TODO: more reliable way of checking for flash attention?
+        if is_nvidia(): #pytorch flash attention only works on Nvidia
+            return True
+    return False
+
+def get_free_memory(dev=None, torch_free_too=False):
+    global directml_enabled
+    if dev is None:
+        dev = get_torch_device()
+
+    if hasattr(dev, 'type') and (dev.type == 'cpu' or dev.type == 'mps'):
+        mem_free_total = psutil.virtual_memory().available
+        mem_free_torch = mem_free_total
+    else:
+        if directml_enabled:
+            mem_free_total = 1024 * 1024 * 1024 #TODO
+            mem_free_torch = mem_free_total
+        elif is_intel_xpu():
+            stats = torch.xpu.memory_stats(dev)
+            mem_active = stats['active_bytes.all.current']
+            mem_allocated = stats['allocated_bytes.all.current']
+            mem_reserved = stats['reserved_bytes.all.current']
+            mem_free_torch = mem_reserved - mem_active
+            mem_free_total = torch.xpu.get_device_properties(dev).total_memory - mem_allocated
+        else:
+            stats = torch.cuda.memory_stats(dev)
+            mem_active = stats['active_bytes.all.current']
+            mem_reserved = stats['reserved_bytes.all.current']
+            mem_free_cuda, _ = torch.cuda.mem_get_info(dev)
+            mem_free_torch = mem_reserved - mem_active
+            mem_free_total = mem_free_cuda + mem_free_torch
+
+    if torch_free_too:
+        return (mem_free_total, mem_free_torch)
+    else:
+        return mem_free_total
+
+def batch_area_memory(area):
+    if xformers_enabled() or pytorch_attention_flash_attention():
+        #TODO: these formulas are copied from maximum_batch_area below
+        return (area / 20) * (1024 * 1024)
+    else:
+        return (((area * 0.6) / 0.9) + 1024) * (1024 * 1024)
+
+def maximum_batch_area():
+    global vram_state
+    if vram_state == VRAMState.NO_VRAM:
+        return 0
+
+    memory_free = get_free_memory() / (1024 * 1024)
+    if xformers_enabled() or pytorch_attention_flash_attention():
+        #TODO: this needs to be tweaked
+        area = 20 * memory_free
+    else:
+        #TODO: this formula is because AMD sucks and has memory management issues which might be fixed in the future
+        area = ((memory_free - 1024) * 0.9) / (0.6)
+    return int(max(area, 0))
+
+def cpu_mode():
+    global cpu_state
+    return cpu_state == CPUState.CPU
+
+def mps_mode():
+    global cpu_state
+    return cpu_state == CPUState.MPS
+
+def is_device_cpu(device):
+    if hasattr(device, 'type'):
+        if (device.type == 'cpu'):
+            return True
+    return False
+
+def is_device_mps(device):
+    if hasattr(device, 'type'):
+        if (device.type == 'mps'):
+            return True
+    return False
+
+def should_use_fp16(device=None, model_params=0, prioritize_performance=True):
+    global directml_enabled
+
+    if device is not None:
+        if is_device_cpu(device):
+            return False
+
+    if FORCE_FP16:
+        return True
+
+    if device is not None: #TODO
+        if is_device_mps(device):
+            return False
+
+    if FORCE_FP32:
+        return False
+
+    if directml_enabled:
+        return False
+
+    if cpu_mode() or mps_mode():
+        return False #TODO ?
+
+    if is_intel_xpu():
+        return True
+
+    if torch.cuda.is_bf16_supported():
+        return True
+
+    props = torch.cuda.get_device_properties("cuda")
+    if props.major < 6:
+        return False
+
+    fp16_works = False
+    #FP16 is confirmed working on a 1080 (GP104) but it's a bit slower than FP32 so it should only be enabled
+    #when the model doesn't actually fit on the card
+    #TODO: actually test if GP106 and others have the same type of behavior
+    nvidia_10_series = ["1080", "1070", "titan x", "p3000", "p3200", "p4000", "p4200", "p5000", "p5200", "p6000", "1060", "1050"]
+    for x in nvidia_10_series:
+        if x in props.name.lower():
+            fp16_works = True
+
+    if fp16_works:
+        free_model_memory = (get_free_memory() * 0.9 - minimum_inference_memory())
+        if (not prioritize_performance) or model_params * 4 > free_model_memory:
+            return True
+
+    if props.major < 7:
+        return False
+
+    #FP16 is just broken on these cards
+    nvidia_16_series = ["1660", "1650", "1630", "T500", "T550", "T600", "MX550", "MX450", "CMP 30HX", "T2000", "T1000", "T1200"]
+    for x in nvidia_16_series:
+        if x in props.name:
+            return False
+
+    return True
+
+def soft_empty_cache(force=False):
+    global cpu_state
+    if cpu_state == CPUState.MPS:
+        torch.mps.empty_cache()
+    elif is_intel_xpu():
+        torch.xpu.empty_cache()
+    elif torch.cuda.is_available():
+        if force or is_nvidia(): #This seems to make things worse on ROCm so I only do it for cuda
+            torch.cuda.empty_cache()
+            torch.cuda.ipc_collect()
+
+def resolve_lowvram_weight(weight, model, key):
+    if weight.device == torch.device("meta"): #lowvram NOTE: this depends on the inner working of the accelerate library so it might break.
+        key_split = key.split('.')              # I have no idea why they don't just leave the weight there instead of using the meta device.
+        op = fcbh.utils.get_attr(model, '.'.join(key_split[:-1]))
+        weight = op._hf_hook.weights_map[key_split[-1]]
+    return weight
+
+#TODO: might be cleaner to put this somewhere else
+import threading
+
+class InterruptProcessingException(Exception):
+    pass
+
+interrupt_processing_mutex = threading.RLock()
+
+interrupt_processing = False
+def interrupt_current_processing(value=True):
+    global interrupt_processing
+    global interrupt_processing_mutex
+    with interrupt_processing_mutex:
+        interrupt_processing = value
+
+def processing_interrupted():
+    global interrupt_processing
+    global interrupt_processing_mutex
+    with interrupt_processing_mutex:
+        return interrupt_processing
+
+def throw_exception_if_processing_interrupted():
+    global interrupt_processing
+    global interrupt_processing_mutex
+    with interrupt_processing_mutex:
+        if interrupt_processing:
+            interrupt_processing = False
+            raise InterruptProcessingException()

backend/headless/fcbh/model_patcher.py

@@ -0,0 +1,311 @@
+import torch
+import copy
+import inspect
+
+import fcbh.utils
+import fcbh.model_management
+
+class ModelPatcher:
+    def __init__(self, model, load_device, offload_device, size=0, current_device=None):
+        self.size = size
+        self.model = model
+        self.patches = {}
+        self.backup = {}
+        self.object_patches = {}
+        self.object_patches_backup = {}
+        self.model_options = {"transformer_options":{}}
+        self.model_size()
+        self.load_device = load_device
+        self.offload_device = offload_device
+        if current_device is None:
+            self.current_device = self.offload_device
+        else:
+            self.current_device = current_device
+
+    def model_size(self):
+        if self.size > 0:
+            return self.size
+        model_sd = self.model.state_dict()
+        size = 0
+        for k in model_sd:
+            t = model_sd[k]
+            size += t.nelement() * t.element_size()
+        self.size = size
+        self.model_keys = set(model_sd.keys())
+        return size
+
+    def clone(self):
+        n = ModelPatcher(self.model, self.load_device, self.offload_device, self.size, self.current_device)
+        n.patches = {}
+        for k in self.patches:
+            n.patches[k] = self.patches[k][:]
+
+        n.object_patches = self.object_patches.copy()
+        n.model_options = copy.deepcopy(self.model_options)
+        n.model_keys = self.model_keys
+        return n
+
+    def is_clone(self, other):
+        if hasattr(other, 'model') and self.model is other.model:
+            return True
+        return False
+
+    def set_model_sampler_cfg_function(self, sampler_cfg_function):
+        if len(inspect.signature(sampler_cfg_function).parameters) == 3:
+            self.model_options["sampler_cfg_function"] = lambda args: sampler_cfg_function(args["cond"], args["uncond"], args["cond_scale"]) #Old way
+        else:
+            self.model_options["sampler_cfg_function"] = sampler_cfg_function
+
+    def set_model_unet_function_wrapper(self, unet_wrapper_function):
+        self.model_options["model_function_wrapper"] = unet_wrapper_function
+
+    def set_model_patch(self, patch, name):
+        to = self.model_options["transformer_options"]
+        if "patches" not in to:
+            to["patches"] = {}
+        to["patches"][name] = to["patches"].get(name, []) + [patch]
+
+    def set_model_patch_replace(self, patch, name, block_name, number):
+        to = self.model_options["transformer_options"]
+        if "patches_replace" not in to:
+            to["patches_replace"] = {}
+        if name not in to["patches_replace"]:
+            to["patches_replace"][name] = {}
+        to["patches_replace"][name][(block_name, number)] = patch
+
+    def set_model_attn1_patch(self, patch):
+        self.set_model_patch(patch, "attn1_patch")
+
+    def set_model_attn2_patch(self, patch):
+        self.set_model_patch(patch, "attn2_patch")
+
+    def set_model_attn1_replace(self, patch, block_name, number):
+        self.set_model_patch_replace(patch, "attn1", block_name, number)
+
+    def set_model_attn2_replace(self, patch, block_name, number):
+        self.set_model_patch_replace(patch, "attn2", block_name, number)
+
+    def set_model_attn1_output_patch(self, patch):
+        self.set_model_patch(patch, "attn1_output_patch")
+
+    def set_model_attn2_output_patch(self, patch):
+        self.set_model_patch(patch, "attn2_output_patch")
+
+    def set_model_output_block_patch(self, patch):
+        self.set_model_patch(patch, "output_block_patch")
+
+    def add_object_patch(self, name, obj):
+        self.object_patches[name] = obj
+
+    def model_patches_to(self, device):
+        to = self.model_options["transformer_options"]
+        if "patches" in to:
+            patches = to["patches"]
+            for name in patches:
+                patch_list = patches[name]
+                for i in range(len(patch_list)):
+                    if hasattr(patch_list[i], "to"):
+                        patch_list[i] = patch_list[i].to(device)
+        if "patches_replace" in to:
+            patches = to["patches_replace"]
+            for name in patches:
+                patch_list = patches[name]
+                for k in patch_list:
+                    if hasattr(patch_list[k], "to"):
+                        patch_list[k] = patch_list[k].to(device)
+        if "model_function_wrapper" in self.model_options:
+            wrap_func = self.model_options["model_function_wrapper"]
+            if hasattr(wrap_func, "to"):
+                self.model_options["model_function_wrapper"] = wrap_func.to(device)
+
+    def model_dtype(self):
+        if hasattr(self.model, "get_dtype"):
+            return self.model.get_dtype()
+
+    def add_patches(self, patches, strength_patch=1.0, strength_model=1.0):
+        p = set()
+        for k in patches:
+            if k in self.model_keys:
+                p.add(k)
+                current_patches = self.patches.get(k, [])
+                current_patches.append((strength_patch, patches[k], strength_model))
+                self.patches[k] = current_patches
+
+        return list(p)
+
+    def get_key_patches(self, filter_prefix=None):
+        fcbh.model_management.unload_model_clones(self)
+        model_sd = self.model_state_dict()
+        p = {}
+        for k in model_sd:
+            if filter_prefix is not None:
+                if not k.startswith(filter_prefix):
+                    continue
+            if k in self.patches:
+                p[k] = [model_sd[k]] + self.patches[k]
+            else:
+                p[k] = (model_sd[k],)
+        return p
+
+    def model_state_dict(self, filter_prefix=None):
+        sd = self.model.state_dict()
+        keys = list(sd.keys())
+        if filter_prefix is not None:
+            for k in keys:
+                if not k.startswith(filter_prefix):
+                    sd.pop(k)
+        return sd
+
+    def patch_model(self, device_to=None):
+        for k in self.object_patches:
+            old = getattr(self.model, k)
+            if k not in self.object_patches_backup:
+                self.object_patches_backup[k] = old
+            setattr(self.model, k, self.object_patches[k])
+
+        model_sd = self.model_state_dict()
+        for key in self.patches:
+            if key not in model_sd:
+                print("could not patch. key doesn't exist in model:", key)
+                continue
+
+            weight = model_sd[key]
+
+            if key not in self.backup:
+                self.backup[key] = weight.to(self.offload_device)
+
+            if device_to is not None:
+                temp_weight = fcbh.model_management.cast_to_device(weight, device_to, torch.float32, copy=True)
+            else:
+                temp_weight = weight.to(torch.float32, copy=True)
+            out_weight = self.calculate_weight(self.patches[key], temp_weight, key).to(weight.dtype)
+            fcbh.utils.set_attr(self.model, key, out_weight)
+            del temp_weight
+
+        if device_to is not None:
+            self.model.to(device_to)
+            self.current_device = device_to
+
+        return self.model
+
+    def calculate_weight(self, patches, weight, key):
+        for p in patches:
+            alpha = p[0]
+            v = p[1]
+            strength_model = p[2]
+
+            if strength_model != 1.0:
+                weight *= strength_model
+
+            if isinstance(v, list):
+                v = (self.calculate_weight(v[1:], v[0].clone(), key), )
+
+            if len(v) == 1:
+                w1 = v[0]
+                if alpha != 0.0:
+                    if w1.shape != weight.shape:
+                        print("WARNING SHAPE MISMATCH {} WEIGHT NOT MERGED {} != {}".format(key, w1.shape, weight.shape))
+                    else:
+                        weight += alpha * fcbh.model_management.cast_to_device(w1, weight.device, weight.dtype)
+            elif len(v) == 4: #lora/locon
+                mat1 = fcbh.model_management.cast_to_device(v[0], weight.device, torch.float32)
+                mat2 = fcbh.model_management.cast_to_device(v[1], weight.device, torch.float32)
+                if v[2] is not None:
+                    alpha *= v[2] / mat2.shape[0]
+                if v[3] is not None:
+                    #locon mid weights, hopefully the math is fine because I didn't properly test it
+                    mat3 = fcbh.model_management.cast_to_device(v[3], weight.device, torch.float32)
+                    final_shape = [mat2.shape[1], mat2.shape[0], mat3.shape[2], mat3.shape[3]]
+                    mat2 = torch.mm(mat2.transpose(0, 1).flatten(start_dim=1), mat3.transpose(0, 1).flatten(start_dim=1)).reshape(final_shape).transpose(0, 1)
+                try:
+                    weight += (alpha * torch.mm(mat1.flatten(start_dim=1), mat2.flatten(start_dim=1))).reshape(weight.shape).type(weight.dtype)
+                except Exception as e:
+                    print("ERROR", key, e)
+            elif len(v) == 8: #lokr
+                w1 = v[0]
+                w2 = v[1]
+                w1_a = v[3]
+                w1_b = v[4]
+                w2_a = v[5]
+                w2_b = v[6]
+                t2 = v[7]
+                dim = None
+
+                if w1 is None:
+                    dim = w1_b.shape[0]
+                    w1 = torch.mm(fcbh.model_management.cast_to_device(w1_a, weight.device, torch.float32),
+                                  fcbh.model_management.cast_to_device(w1_b, weight.device, torch.float32))
+                else:
+                    w1 = fcbh.model_management.cast_to_device(w1, weight.device, torch.float32)
+
+                if w2 is None:
+                    dim = w2_b.shape[0]
+                    if t2 is None:
+                        w2 = torch.mm(fcbh.model_management.cast_to_device(w2_a, weight.device, torch.float32),
+                                      fcbh.model_management.cast_to_device(w2_b, weight.device, torch.float32))
+                    else:
+                        w2 = torch.einsum('i j k l, j r, i p -> p r k l',
+                                          fcbh.model_management.cast_to_device(t2, weight.device, torch.float32),
+                                          fcbh.model_management.cast_to_device(w2_b, weight.device, torch.float32),
+                                          fcbh.model_management.cast_to_device(w2_a, weight.device, torch.float32))
+                else:
+                    w2 = fcbh.model_management.cast_to_device(w2, weight.device, torch.float32)
+
+                if len(w2.shape) == 4:
+                    w1 = w1.unsqueeze(2).unsqueeze(2)
+                if v[2] is not None and dim is not None:
+                    alpha *= v[2] / dim
+
+                try:
+                    weight += alpha * torch.kron(w1, w2).reshape(weight.shape).type(weight.dtype)
+                except Exception as e:
+                    print("ERROR", key, e)
+            else: #loha
+                w1a = v[0]
+                w1b = v[1]
+                if v[2] is not None:
+                    alpha *= v[2] / w1b.shape[0]
+                w2a = v[3]
+                w2b = v[4]
+                if v[5] is not None: #cp decomposition
+                    t1 = v[5]
+                    t2 = v[6]
+                    m1 = torch.einsum('i j k l, j r, i p -> p r k l',
+                                      fcbh.model_management.cast_to_device(t1, weight.device, torch.float32),
+                                      fcbh.model_management.cast_to_device(w1b, weight.device, torch.float32),
+                                      fcbh.model_management.cast_to_device(w1a, weight.device, torch.float32))
+
+                    m2 = torch.einsum('i j k l, j r, i p -> p r k l',
+                                      fcbh.model_management.cast_to_device(t2, weight.device, torch.float32),
+                                      fcbh.model_management.cast_to_device(w2b, weight.device, torch.float32),
+                                      fcbh.model_management.cast_to_device(w2a, weight.device, torch.float32))
+                else:
+                    m1 = torch.mm(fcbh.model_management.cast_to_device(w1a, weight.device, torch.float32),
+                                  fcbh.model_management.cast_to_device(w1b, weight.device, torch.float32))
+                    m2 = torch.mm(fcbh.model_management.cast_to_device(w2a, weight.device, torch.float32),
+                                  fcbh.model_management.cast_to_device(w2b, weight.device, torch.float32))
+
+                try:
+                    weight += (alpha * m1 * m2).reshape(weight.shape).type(weight.dtype)
+                except Exception as e:
+                    print("ERROR", key, e)
+
+        return weight
+
+    def unpatch_model(self, device_to=None):
+        keys = list(self.backup.keys())
+
+        for k in keys:
+            fcbh.utils.set_attr(self.model, k, self.backup[k])
+
+        self.backup = {}
+
+        if device_to is not None:
+            self.model.to(device_to)
+            self.current_device = device_to
+
+        keys = list(self.object_patches_backup.keys())
+        for k in keys:
+            setattr(self.model, k, self.object_patches_backup[k])
+
+        self.object_patches_backup = {}

backend/headless/fcbh/model_sampling.py

@@ -0,0 +1,80 @@
+import torch
+import numpy as np
+from fcbh.ldm.modules.diffusionmodules.util import make_beta_schedule
+
+
+class EPS:
+    def calculate_input(self, sigma, noise):
+        sigma = sigma.view(sigma.shape[:1] + (1,) * (noise.ndim - 1))
+        return noise / (sigma ** 2 + self.sigma_data ** 2) ** 0.5
+
+    def calculate_denoised(self, sigma, model_output, model_input):
+        sigma = sigma.view(sigma.shape[:1] + (1,) * (model_output.ndim - 1))
+        return model_input - model_output * sigma
+
+
+class V_PREDICTION(EPS):
+    def calculate_denoised(self, sigma, model_output, model_input):
+        sigma = sigma.view(sigma.shape[:1] + (1,) * (model_output.ndim - 1))
+        return model_input * self.sigma_data ** 2 / (sigma ** 2 + self.sigma_data ** 2) - model_output * sigma * self.sigma_data / (sigma ** 2 + self.sigma_data ** 2) ** 0.5
+
+
+class ModelSamplingDiscrete(torch.nn.Module):
+    def __init__(self, model_config=None):
+        super().__init__()
+        beta_schedule = "linear"
+        if model_config is not None:
+            beta_schedule = model_config.beta_schedule
+        self._register_schedule(given_betas=None, beta_schedule=beta_schedule, timesteps=1000, linear_start=0.00085, linear_end=0.012, cosine_s=8e-3)
+        self.sigma_data = 1.0
+
+    def _register_schedule(self, given_betas=None, beta_schedule="linear", timesteps=1000,
+                          linear_start=1e-4, linear_end=2e-2, cosine_s=8e-3):
+        if given_betas is not None:
+            betas = given_betas
+        else:
+            betas = make_beta_schedule(beta_schedule, timesteps, linear_start=linear_start, linear_end=linear_end, cosine_s=cosine_s)
+        alphas = 1. - betas
+        alphas_cumprod = torch.tensor(np.cumprod(alphas, axis=0), dtype=torch.float32)
+        # alphas_cumprod_prev = np.append(1., alphas_cumprod[:-1])
+
+        timesteps, = betas.shape
+        self.num_timesteps = int(timesteps)
+        self.linear_start = linear_start
+        self.linear_end = linear_end
+
+        # self.register_buffer('betas', torch.tensor(betas, dtype=torch.float32))
+        # self.register_buffer('alphas_cumprod', torch.tensor(alphas_cumprod, dtype=torch.float32))
+        # self.register_buffer('alphas_cumprod_prev', torch.tensor(alphas_cumprod_prev, dtype=torch.float32))
+
+        sigmas = ((1 - alphas_cumprod) / alphas_cumprod) ** 0.5
+        self.set_sigmas(sigmas)
+
+    def set_sigmas(self, sigmas):
+        self.register_buffer('sigmas', sigmas)
+        self.register_buffer('log_sigmas', sigmas.log())
+
+    @property
+    def sigma_min(self):
+        return self.sigmas[0]
+
+    @property
+    def sigma_max(self):
+        return self.sigmas[-1]
+
+    def timestep(self, sigma):
+        log_sigma = sigma.log()
+        dists = log_sigma.to(self.log_sigmas.device) - self.log_sigmas[:, None]
+        return dists.abs().argmin(dim=0).view(sigma.shape)
+
+    def sigma(self, timestep):
+        t = torch.clamp(timestep.float(), min=0, max=(len(self.sigmas) - 1))
+        low_idx = t.floor().long()
+        high_idx = t.ceil().long()
+        w = t.frac()
+        log_sigma = (1 - w) * self.log_sigmas[low_idx] + w * self.log_sigmas[high_idx]
+        return log_sigma.exp()
+
+    def percent_to_sigma(self, percent):
+        return self.sigma(torch.tensor(percent * 999.0))
+

backend/headless/fcbh/ops.py

@@ -0,0 +1,40 @@
+import torch
+from contextlib import contextmanager
+
+class Linear(torch.nn.Linear):
+    def reset_parameters(self):
+        return None
+
+class Conv2d(torch.nn.Conv2d):
+    def reset_parameters(self):
+        return None
+
+class Conv3d(torch.nn.Conv3d):
+    def reset_parameters(self):
+        return None
+
+def conv_nd(dims, *args, **kwargs):
+    if dims == 2:
+        return Conv2d(*args, **kwargs)
+    elif dims == 3:
+        return Conv3d(*args, **kwargs)
+    else:
+        raise ValueError(f"unsupported dimensions: {dims}")
+
+@contextmanager
+def use_fcbh_ops(device=None, dtype=None): # Kind of an ugly hack but I can't think of a better way
+    old_torch_nn_linear = torch.nn.Linear
+    force_device = device
+    force_dtype = dtype
+    def linear_with_dtype(in_features: int, out_features: int, bias: bool = True, device=None, dtype=None):
+        if force_device is not None:
+            device = force_device
+        if force_dtype is not None:
+            dtype = force_dtype
+        return Linear(in_features, out_features, bias=bias, device=device, dtype=dtype)
+
+    torch.nn.Linear = linear_with_dtype
+    try:
+        yield
+    finally:
+        torch.nn.Linear = old_torch_nn_linear

backend/headless/fcbh/options.py

@@ -0,0 +1,6 @@
+
+args_parsing = False
+
+def enable_args_parsing(enable=True):
+    global args_parsing
+    args_parsing = enable

backend/headless/fcbh/sample.py

@@ -0,0 +1,118 @@
+import torch
+import fcbh.model_management
+import fcbh.samplers
+import fcbh.conds
+import fcbh.utils
+import math
+import numpy as np
+
+def prepare_noise(latent_image, seed, noise_inds=None):
+    """
+    creates random noise given a latent image and a seed.
+    optional arg skip can be used to skip and discard x number of noise generations for a given seed
+    """
+    generator = torch.manual_seed(seed)
+    if noise_inds is None:
+        return torch.randn(latent_image.size(), dtype=latent_image.dtype, layout=latent_image.layout, generator=generator, device="cpu")
+    
+    unique_inds, inverse = np.unique(noise_inds, return_inverse=True)
+    noises = []
+    for i in range(unique_inds[-1]+1):
+        noise = torch.randn([1] + list(latent_image.size())[1:], dtype=latent_image.dtype, layout=latent_image.layout, generator=generator, device="cpu")
+        if i in unique_inds:
+            noises.append(noise)
+    noises = [noises[i] for i in inverse]
+    noises = torch.cat(noises, axis=0)
+    return noises
+
+def prepare_mask(noise_mask, shape, device):
+    """ensures noise mask is of proper dimensions"""
+    noise_mask = torch.nn.functional.interpolate(noise_mask.reshape((-1, 1, noise_mask.shape[-2], noise_mask.shape[-1])), size=(shape[2], shape[3]), mode="bilinear")
+    noise_mask = noise_mask.round()
+    noise_mask = torch.cat([noise_mask] * shape[1], dim=1)
+    noise_mask = fcbh.utils.repeat_to_batch_size(noise_mask, shape[0])
+    noise_mask = noise_mask.to(device)
+    return noise_mask
+
+def get_models_from_cond(cond, model_type):
+    models = []
+    for c in cond:
+        if model_type in c:
+            models += [c[model_type]]
+    return models
+
+def convert_cond(cond):
+    out = []
+    for c in cond:
+        temp = c[1].copy()
+        model_conds = temp.get("model_conds", {})
+        if c[0] is not None:
+            model_conds["c_crossattn"] = fcbh.conds.CONDCrossAttn(c[0])
+        temp["model_conds"] = model_conds
+        out.append(temp)
+    return out
+
+def get_additional_models(positive, negative, dtype):
+    """loads additional models in positive and negative conditioning"""
+    control_nets = set(get_models_from_cond(positive, "control") + get_models_from_cond(negative, "control"))
+
+    inference_memory = 0
+    control_models = []
+    for m in control_nets:
+        control_models += m.get_models()
+        inference_memory += m.inference_memory_requirements(dtype)
+
+    gligen = get_models_from_cond(positive, "gligen") + get_models_from_cond(negative, "gligen")
+    gligen = [x[1] for x in gligen]
+    models = control_models + gligen
+    return models, inference_memory
+
+def cleanup_additional_models(models):
+    """cleanup additional models that were loaded"""
+    for m in models:
+        if hasattr(m, 'cleanup'):
+            m.cleanup()
+
+def prepare_sampling(model, noise_shape, positive, negative, noise_mask):
+    device = model.load_device
+    positive = convert_cond(positive)
+    negative = convert_cond(negative)
+
+    if noise_mask is not None:
+        noise_mask = prepare_mask(noise_mask, noise_shape, device)
+
+    real_model = None
+    models, inference_memory = get_additional_models(positive, negative, model.model_dtype())
+    fcbh.model_management.load_models_gpu([model] + models, fcbh.model_management.batch_area_memory(noise_shape[0] * noise_shape[2] * noise_shape[3]) + inference_memory)
+    real_model = model.model
+
+    return real_model, positive, negative, noise_mask, models
+
+
+def sample(model, noise, steps, cfg, sampler_name, scheduler, positive, negative, latent_image, denoise=1.0, disable_noise=False, start_step=None, last_step=None, force_full_denoise=False, noise_mask=None, sigmas=None, callback=None, disable_pbar=False, seed=None):
+    real_model, positive_copy, negative_copy, noise_mask, models = prepare_sampling(model, noise.shape, positive, negative, noise_mask)
+
+    noise = noise.to(model.load_device)
+    latent_image = latent_image.to(model.load_device)
+
+    sampler = fcbh.samplers.KSampler(real_model, steps=steps, device=model.load_device, sampler=sampler_name, scheduler=scheduler, denoise=denoise, model_options=model.model_options)
+
+    samples = sampler.sample(noise, positive_copy, negative_copy, cfg=cfg, latent_image=latent_image, start_step=start_step, last_step=last_step, force_full_denoise=force_full_denoise, denoise_mask=noise_mask, sigmas=sigmas, callback=callback, disable_pbar=disable_pbar, seed=seed)
+    samples = samples.cpu()
+
+    cleanup_additional_models(models)
+    cleanup_additional_models(set(get_models_from_cond(positive, "control") + get_models_from_cond(negative, "control")))
+    return samples
+
+def sample_custom(model, noise, cfg, sampler, sigmas, positive, negative, latent_image, noise_mask=None, callback=None, disable_pbar=False, seed=None):
+    real_model, positive_copy, negative_copy, noise_mask, models = prepare_sampling(model, noise.shape, positive, negative, noise_mask)
+    noise = noise.to(model.load_device)
+    latent_image = latent_image.to(model.load_device)
+    sigmas = sigmas.to(model.load_device)
+
+    samples = fcbh.samplers.sample(real_model, noise, positive_copy, negative_copy, cfg, model.load_device, sampler, sigmas, model_options=model.model_options, latent_image=latent_image, denoise_mask=noise_mask, callback=callback, disable_pbar=disable_pbar, seed=seed)
+    samples = samples.cpu()
+    cleanup_additional_models(models)
+    cleanup_additional_models(set(get_models_from_cond(positive, "control") + get_models_from_cond(negative, "control")))
+    return samples
+

backend/headless/fcbh/samplers.py

@@ -0,0 +1,691 @@
+from .k_diffusion import sampling as k_diffusion_sampling
+from .extra_samplers import uni_pc
+import torch
+import enum
+from fcbh import model_management
+import math
+from fcbh import model_base
+import fcbh.utils
+import fcbh.conds
+
+
+#The main sampling function shared by all the samplers
+#Returns denoised
+def sampling_function(model_function, x, timestep, uncond, cond, cond_scale, model_options={}, seed=None):
+        def get_area_and_mult(conds, x_in, timestep_in):
+            area = (x_in.shape[2], x_in.shape[3], 0, 0)
+            strength = 1.0
+
+            if 'timestep_start' in conds:
+                timestep_start = conds['timestep_start']
+                if timestep_in[0] > timestep_start:
+                    return None
+            if 'timestep_end' in conds:
+                timestep_end = conds['timestep_end']
+                if timestep_in[0] < timestep_end:
+                    return None
+            if 'area' in conds:
+                area = conds['area']
+            if 'strength' in conds:
+                strength = conds['strength']
+
+            input_x = x_in[:,:,area[2]:area[0] + area[2],area[3]:area[1] + area[3]]
+            if 'mask' in conds:
+                # Scale the mask to the size of the input
+                # The mask should have been resized as we began the sampling process
+                mask_strength = 1.0
+                if "mask_strength" in conds:
+                    mask_strength = conds["mask_strength"]
+                mask = conds['mask']
+                assert(mask.shape[1] == x_in.shape[2])
+                assert(mask.shape[2] == x_in.shape[3])
+                mask = mask[:,area[2]:area[0] + area[2],area[3]:area[1] + area[3]] * mask_strength
+                mask = mask.unsqueeze(1).repeat(input_x.shape[0] // mask.shape[0], input_x.shape[1], 1, 1)
+            else:
+                mask = torch.ones_like(input_x)
+            mult = mask * strength
+
+            if 'mask' not in conds:
+                rr = 8
+                if area[2] != 0:
+                    for t in range(rr):
+                        mult[:,:,t:1+t,:] *= ((1.0/rr) * (t + 1))
+                if (area[0] + area[2]) < x_in.shape[2]:
+                    for t in range(rr):
+                        mult[:,:,area[0] - 1 - t:area[0] - t,:] *= ((1.0/rr) * (t + 1))
+                if area[3] != 0:
+                    for t in range(rr):
+                        mult[:,:,:,t:1+t] *= ((1.0/rr) * (t + 1))
+                if (area[1] + area[3]) < x_in.shape[3]:
+                    for t in range(rr):
+                        mult[:,:,:,area[1] - 1 - t:area[1] - t] *= ((1.0/rr) * (t + 1))
+
+            conditionning = {}
+            model_conds = conds["model_conds"]
+            for c in model_conds:
+                conditionning[c] = model_conds[c].process_cond(batch_size=x_in.shape[0], device=x_in.device, area=area)
+
+            control = None
+            if 'control' in conds:
+                control = conds['control']
+
+            patches = None
+            if 'gligen' in conds:
+                gligen = conds['gligen']
+                patches = {}
+                gligen_type = gligen[0]
+                gligen_model = gligen[1]
+                if gligen_type == "position":
+                    gligen_patch = gligen_model.model.set_position(input_x.shape, gligen[2], input_x.device)
+                else:
+                    gligen_patch = gligen_model.model.set_empty(input_x.shape, input_x.device)
+
+                patches['middle_patch'] = [gligen_patch]
+
+            return (input_x, mult, conditionning, area, control, patches)
+
+        def cond_equal_size(c1, c2):
+            if c1 is c2:
+                return True
+            if c1.keys() != c2.keys():
+                return False
+            for k in c1:
+                if not c1[k].can_concat(c2[k]):
+                    return False
+            return True
+
+        def can_concat_cond(c1, c2):
+            if c1[0].shape != c2[0].shape:
+                return False
+
+            #control
+            if (c1[4] is None) != (c2[4] is None):
+                return False
+            if c1[4] is not None:
+                if c1[4] is not c2[4]:
+                    return False
+
+            #patches
+            if (c1[5] is None) != (c2[5] is None):
+                return False
+            if (c1[5] is not None):
+                if c1[5] is not c2[5]:
+                    return False
+
+            return cond_equal_size(c1[2], c2[2])
+
+        def cond_cat(c_list):
+            c_crossattn = []
+            c_concat = []
+            c_adm = []
+            crossattn_max_len = 0
+
+            temp = {}
+            for x in c_list:
+                for k in x:
+                    cur = temp.get(k, [])
+                    cur.append(x[k])
+                    temp[k] = cur
+
+            out = {}
+            for k in temp:
+                conds = temp[k]
+                out[k] = conds[0].concat(conds[1:])
+
+            return out
+
+        def calc_cond_uncond_batch(model_function, cond, uncond, x_in, timestep, max_total_area, model_options):
+            out_cond = torch.zeros_like(x_in)
+            out_count = torch.ones_like(x_in) * 1e-37
+
+            out_uncond = torch.zeros_like(x_in)
+            out_uncond_count = torch.ones_like(x_in) * 1e-37
+
+            COND = 0
+            UNCOND = 1
+
+            to_run = []
+            for x in cond:
+                p = get_area_and_mult(x, x_in, timestep)
+                if p is None:
+                    continue
+
+                to_run += [(p, COND)]
+            if uncond is not None:
+                for x in uncond:
+                    p = get_area_and_mult(x, x_in, timestep)
+                    if p is None:
+                        continue
+
+                    to_run += [(p, UNCOND)]
+
+            while len(to_run) > 0:
+                first = to_run[0]
+                first_shape = first[0][0].shape
+                to_batch_temp = []
+                for x in range(len(to_run)):
+                    if can_concat_cond(to_run[x][0], first[0]):
+                        to_batch_temp += [x]
+
+                to_batch_temp.reverse()
+                to_batch = to_batch_temp[:1]
+
+                for i in range(1, len(to_batch_temp) + 1):
+                    batch_amount = to_batch_temp[:len(to_batch_temp)//i]
+                    if (len(batch_amount) * first_shape[0] * first_shape[2] * first_shape[3] < max_total_area):
+                        to_batch = batch_amount
+                        break
+
+                input_x = []
+                mult = []
+                c = []
+                cond_or_uncond = []
+                area = []
+                control = None
+                patches = None
+                for x in to_batch:
+                    o = to_run.pop(x)
+                    p = o[0]
+                    input_x += [p[0]]
+                    mult += [p[1]]
+                    c += [p[2]]
+                    area += [p[3]]
+                    cond_or_uncond += [o[1]]
+                    control = p[4]
+                    patches = p[5]
+
+                batch_chunks = len(cond_or_uncond)
+                input_x = torch.cat(input_x)
+                c = cond_cat(c)
+                timestep_ = torch.cat([timestep] * batch_chunks)
+
+                if control is not None:
+                    c['control'] = control.get_control(input_x, timestep_, c, len(cond_or_uncond))
+
+                transformer_options = {}
+                if 'transformer_options' in model_options:
+                    transformer_options = model_options['transformer_options'].copy()
+
+                if patches is not None:
+                    if "patches" in transformer_options:
+                        cur_patches = transformer_options["patches"].copy()
+                        for p in patches:
+                            if p in cur_patches:
+                                cur_patches[p] = cur_patches[p] + patches[p]
+                            else:
+                                cur_patches[p] = patches[p]
+                    else:
+                        transformer_options["patches"] = patches
+
+                transformer_options["cond_or_uncond"] = cond_or_uncond[:]
+                c['transformer_options'] = transformer_options
+
+                if 'model_function_wrapper' in model_options:
+                    output = model_options['model_function_wrapper'](model_function, {"input": input_x, "timestep": timestep_, "c": c, "cond_or_uncond": cond_or_uncond}).chunk(batch_chunks)
+                else:
+                    output = model_function(input_x, timestep_, **c).chunk(batch_chunks)
+                del input_x
+
+                for o in range(batch_chunks):
+                    if cond_or_uncond[o] == COND:
+                        out_cond[:,:,area[o][2]:area[o][0] + area[o][2],area[o][3]:area[o][1] + area[o][3]] += output[o] * mult[o]
+                        out_count[:,:,area[o][2]:area[o][0] + area[o][2],area[o][3]:area[o][1] + area[o][3]] += mult[o]
+                    else:
+                        out_uncond[:,:,area[o][2]:area[o][0] + area[o][2],area[o][3]:area[o][1] + area[o][3]] += output[o] * mult[o]
+                        out_uncond_count[:,:,area[o][2]:area[o][0] + area[o][2],area[o][3]:area[o][1] + area[o][3]] += mult[o]
+                del mult
+
+            out_cond /= out_count
+            del out_count
+            out_uncond /= out_uncond_count
+            del out_uncond_count
+            return out_cond, out_uncond
+
+
+        max_total_area = model_management.maximum_batch_area()
+        if math.isclose(cond_scale, 1.0):
+            uncond = None
+
+        cond, uncond = calc_cond_uncond_batch(model_function, cond, uncond, x, timestep, max_total_area, model_options)
+        if "sampler_cfg_function" in model_options:
+            args = {"cond": x - cond, "uncond": x - uncond, "cond_scale": cond_scale, "timestep": timestep, "input": x, "sigma": timestep}
+            return x - model_options["sampler_cfg_function"](args)
+        else:
+            return uncond + (cond - uncond) * cond_scale
+
+class CFGNoisePredictor(torch.nn.Module):
+    def __init__(self, model):
+        super().__init__()
+        self.inner_model = model
+    def apply_model(self, x, timestep, cond, uncond, cond_scale, model_options={}, seed=None):
+        out = sampling_function(self.inner_model.apply_model, x, timestep, uncond, cond, cond_scale, model_options=model_options, seed=seed)
+        return out
+    def forward(self, *args, **kwargs):
+        return self.apply_model(*args, **kwargs)
+
+class KSamplerX0Inpaint(torch.nn.Module):
+    def __init__(self, model):
+        super().__init__()
+        self.inner_model = model
+    def forward(self, x, sigma, uncond, cond, cond_scale, denoise_mask, model_options={}, seed=None):
+        if denoise_mask is not None:
+            latent_mask = 1. - denoise_mask
+            x = x * denoise_mask + (self.latent_image + self.noise * sigma.reshape([sigma.shape[0]] + [1] * (len(self.noise.shape) - 1))) * latent_mask
+        out = self.inner_model(x, sigma, cond=cond, uncond=uncond, cond_scale=cond_scale, model_options=model_options, seed=seed)
+        if denoise_mask is not None:
+            out *= denoise_mask
+
+        if denoise_mask is not None:
+            out += self.latent_image * latent_mask
+        return out
+
+def simple_scheduler(model, steps):
+    s = model.model_sampling
+    sigs = []
+    ss = len(s.sigmas) / steps
+    for x in range(steps):
+        sigs += [float(s.sigmas[-(1 + int(x * ss))])]
+    sigs += [0.0]
+    return torch.FloatTensor(sigs)
+
+def ddim_scheduler(model, steps):
+    s = model.model_sampling
+    sigs = []
+    ss = len(s.sigmas) // steps
+    x = 1
+    while x < len(s.sigmas):
+        sigs += [float(s.sigmas[x])]
+        x += ss
+    sigs = sigs[::-1]
+    sigs += [0.0]
+    return torch.FloatTensor(sigs)
+
+def normal_scheduler(model, steps, sgm=False, floor=False):
+    s = model.model_sampling
+    start = s.timestep(s.sigma_max)
+    end = s.timestep(s.sigma_min)
+
+    if sgm:
+        timesteps = torch.linspace(start, end, steps + 1)[:-1]
+    else:
+        timesteps = torch.linspace(start, end, steps)
+
+    sigs = []
+    for x in range(len(timesteps)):
+        ts = timesteps[x]
+        sigs.append(s.sigma(ts))
+    sigs += [0.0]
+    return torch.FloatTensor(sigs)
+
+def get_mask_aabb(masks):
+    if masks.numel() == 0:
+        return torch.zeros((0, 4), device=masks.device, dtype=torch.int)
+
+    b = masks.shape[0]
+
+    bounding_boxes = torch.zeros((b, 4), device=masks.device, dtype=torch.int)
+    is_empty = torch.zeros((b), device=masks.device, dtype=torch.bool)
+    for i in range(b):
+        mask = masks[i]
+        if mask.numel() == 0:
+            continue
+        if torch.max(mask != 0) == False:
+            is_empty[i] = True
+            continue
+        y, x = torch.where(mask)
+        bounding_boxes[i, 0] = torch.min(x)
+        bounding_boxes[i, 1] = torch.min(y)
+        bounding_boxes[i, 2] = torch.max(x)
+        bounding_boxes[i, 3] = torch.max(y)
+
+    return bounding_boxes, is_empty
+
+def resolve_areas_and_cond_masks(conditions, h, w, device):
+    # We need to decide on an area outside the sampling loop in order to properly generate opposite areas of equal sizes.
+    # While we're doing this, we can also resolve the mask device and scaling for performance reasons
+    for i in range(len(conditions)):
+        c = conditions[i]
+        if 'area' in c:
+            area = c['area']
+            if area[0] == "percentage":
+                modified = c.copy()
+                area = (max(1, round(area[1] * h)), max(1, round(area[2] * w)), round(area[3] * h), round(area[4] * w))
+                modified['area'] = area
+                c = modified
+                conditions[i] = c
+
+        if 'mask' in c:
+            mask = c['mask']
+            mask = mask.to(device=device)
+            modified = c.copy()
+            if len(mask.shape) == 2:
+                mask = mask.unsqueeze(0)
+            if mask.shape[1] != h or mask.shape[2] != w:
+                mask = torch.nn.functional.interpolate(mask.unsqueeze(1), size=(h, w), mode='bilinear', align_corners=False).squeeze(1)
+
+            if modified.get("set_area_to_bounds", False):
+                bounds = torch.max(torch.abs(mask),dim=0).values.unsqueeze(0)
+                boxes, is_empty = get_mask_aabb(bounds)
+                if is_empty[0]:
+                    # Use the minimum possible size for efficiency reasons. (Since the mask is all-0, this becomes a noop anyway)
+                    modified['area'] = (8, 8, 0, 0)
+                else:
+                    box = boxes[0]
+                    H, W, Y, X = (box[3] - box[1] + 1, box[2] - box[0] + 1, box[1], box[0])
+                    H = max(8, H)
+                    W = max(8, W)
+                    area = (int(H), int(W), int(Y), int(X))
+                    modified['area'] = area
+
+            modified['mask'] = mask
+            conditions[i] = modified
+
+def create_cond_with_same_area_if_none(conds, c):
+    if 'area' not in c:
+        return
+
+    c_area = c['area']
+    smallest = None
+    for x in conds:
+        if 'area' in x:
+            a = x['area']
+            if c_area[2] >= a[2] and c_area[3] >= a[3]:
+                if a[0] + a[2] >= c_area[0] + c_area[2]:
+                    if a[1] + a[3] >= c_area[1] + c_area[3]:
+                        if smallest is None:
+                            smallest = x
+                        elif 'area' not in smallest:
+                            smallest = x
+                        else:
+                            if smallest['area'][0] * smallest['area'][1] > a[0] * a[1]:
+                                smallest = x
+        else:
+            if smallest is None:
+                smallest = x
+    if smallest is None:
+        return
+    if 'area' in smallest:
+        if smallest['area'] == c_area:
+            return
+
+    out = c.copy()
+    out['model_conds'] = smallest['model_conds'].copy() #TODO: which fields should be copied?
+    conds += [out]
+
+def calculate_start_end_timesteps(model, conds):
+    s = model.model_sampling
+    for t in range(len(conds)):
+        x = conds[t]
+
+        timestep_start = None
+        timestep_end = None
+        if 'start_percent' in x:
+            timestep_start = s.percent_to_sigma(x['start_percent'])
+        if 'end_percent' in x:
+            timestep_end = s.percent_to_sigma(x['end_percent'])
+
+        if (timestep_start is not None) or (timestep_end is not None):
+            n = x.copy()
+            if (timestep_start is not None):
+                n['timestep_start'] = timestep_start
+            if (timestep_end is not None):
+                n['timestep_end'] = timestep_end
+            conds[t] = n
+
+def pre_run_control(model, conds):
+    s = model.model_sampling
+    for t in range(len(conds)):
+        x = conds[t]
+
+        timestep_start = None
+        timestep_end = None
+        percent_to_timestep_function = lambda a: s.percent_to_sigma(a)
+        if 'control' in x:
+            x['control'].pre_run(model, percent_to_timestep_function)
+
+def apply_empty_x_to_equal_area(conds, uncond, name, uncond_fill_func):
+    cond_cnets = []
+    cond_other = []
+    uncond_cnets = []
+    uncond_other = []
+    for t in range(len(conds)):
+        x = conds[t]
+        if 'area' not in x:
+            if name in x and x[name] is not None:
+                cond_cnets.append(x[name])
+            else:
+                cond_other.append((x, t))
+    for t in range(len(uncond)):
+        x = uncond[t]
+        if 'area' not in x:
+            if name in x and x[name] is not None:
+                uncond_cnets.append(x[name])
+            else:
+                uncond_other.append((x, t))
+
+    if len(uncond_cnets) > 0:
+        return
+
+    for x in range(len(cond_cnets)):
+        temp = uncond_other[x % len(uncond_other)]
+        o = temp[0]
+        if name in o and o[name] is not None:
+            n = o.copy()
+            n[name] = uncond_fill_func(cond_cnets, x)
+            uncond += [n]
+        else:
+            n = o.copy()
+            n[name] = uncond_fill_func(cond_cnets, x)
+            uncond[temp[1]] = n
+
+def encode_model_conds(model_function, conds, noise, device, prompt_type, **kwargs):
+    for t in range(len(conds)):
+        x = conds[t]
+        params = x.copy()
+        params["device"] = device
+        params["noise"] = noise
+        params["width"] = params.get("width", noise.shape[3] * 8)
+        params["height"] = params.get("height", noise.shape[2] * 8)
+        params["prompt_type"] = params.get("prompt_type", prompt_type)
+        for k in kwargs:
+            if k not in params:
+                params[k] = kwargs[k]
+
+        out = model_function(**params)
+        x = x.copy()
+        model_conds = x['model_conds'].copy()
+        for k in out:
+            model_conds[k] = out[k]
+        x['model_conds'] = model_conds
+        conds[t] = x
+    return conds
+
+class Sampler:
+    def sample(self):
+        pass
+
+    def max_denoise(self, model_wrap, sigmas):
+        max_sigma = float(model_wrap.inner_model.model_sampling.sigma_max)
+        sigma = float(sigmas[0])
+        return math.isclose(max_sigma, sigma, rel_tol=1e-05) or sigma > max_sigma
+
+class UNIPC(Sampler):
+    def sample(self, model_wrap, sigmas, extra_args, callback, noise, latent_image=None, denoise_mask=None, disable_pbar=False):
+        return uni_pc.sample_unipc(model_wrap, noise, latent_image, sigmas, sampling_function=sampling_function, max_denoise=self.max_denoise(model_wrap, sigmas), extra_args=extra_args, noise_mask=denoise_mask, callback=callback, disable=disable_pbar)
+
+class UNIPCBH2(Sampler):
+    def sample(self, model_wrap, sigmas, extra_args, callback, noise, latent_image=None, denoise_mask=None, disable_pbar=False):
+        return uni_pc.sample_unipc(model_wrap, noise, latent_image, sigmas, sampling_function=sampling_function, max_denoise=self.max_denoise(model_wrap, sigmas), extra_args=extra_args, noise_mask=denoise_mask, callback=callback, variant='bh2', disable=disable_pbar)
+
+KSAMPLER_NAMES = ["euler", "euler_ancestral", "heun", "dpm_2", "dpm_2_ancestral",
+                  "lms", "dpm_fast", "dpm_adaptive", "dpmpp_2s_ancestral", "dpmpp_sde", "dpmpp_sde_gpu",
+                  "dpmpp_2m", "dpmpp_2m_sde", "dpmpp_2m_sde_gpu", "dpmpp_3m_sde", "dpmpp_3m_sde_gpu", "ddpm", "lcm"]
+
+def ksampler(sampler_name, extra_options={}, inpaint_options={}):
+    class KSAMPLER(Sampler):
+        def sample(self, model_wrap, sigmas, extra_args, callback, noise, latent_image=None, denoise_mask=None, disable_pbar=False):
+            extra_args["denoise_mask"] = denoise_mask
+            model_k = KSamplerX0Inpaint(model_wrap)
+            model_k.latent_image = latent_image
+            if inpaint_options.get("random", False): #TODO: Should this be the default?
+                generator = torch.manual_seed(extra_args.get("seed", 41) + 1)
+                model_k.noise = torch.randn(noise.shape, generator=generator, device="cpu").to(noise.dtype).to(noise.device)
+            else:
+                model_k.noise = noise
+
+            if self.max_denoise(model_wrap, sigmas):
+                noise = noise * torch.sqrt(1.0 + sigmas[0] ** 2.0)
+            else:
+                noise = noise * sigmas[0]
+
+            k_callback = None
+            total_steps = len(sigmas) - 1
+            if callback is not None:
+                k_callback = lambda x: callback(x["i"], x["denoised"], x["x"], total_steps)
+
+            sigma_min = sigmas[-1]
+            if sigma_min == 0:
+                sigma_min = sigmas[-2]
+
+            if latent_image is not None:
+                noise += latent_image
+            if sampler_name == "dpm_fast":
+                samples = k_diffusion_sampling.sample_dpm_fast(model_k, noise, sigma_min, sigmas[0], total_steps, extra_args=extra_args, callback=k_callback, disable=disable_pbar)
+            elif sampler_name == "dpm_adaptive":
+                samples = k_diffusion_sampling.sample_dpm_adaptive(model_k, noise, sigma_min, sigmas[0], extra_args=extra_args, callback=k_callback, disable=disable_pbar)
+            else:
+                samples = getattr(k_diffusion_sampling, "sample_{}".format(sampler_name))(model_k, noise, sigmas, extra_args=extra_args, callback=k_callback, disable=disable_pbar, **extra_options)
+            return samples
+    return KSAMPLER
+
+def wrap_model(model):
+    model_denoise = CFGNoisePredictor(model)
+    return model_denoise
+
+def sample(model, noise, positive, negative, cfg, device, sampler, sigmas, model_options={}, latent_image=None, denoise_mask=None, callback=None, disable_pbar=False, seed=None):
+    positive = positive[:]
+    negative = negative[:]
+
+    resolve_areas_and_cond_masks(positive, noise.shape[2], noise.shape[3], device)
+    resolve_areas_and_cond_masks(negative, noise.shape[2], noise.shape[3], device)
+
+    model_wrap = wrap_model(model)
+
+    calculate_start_end_timesteps(model, negative)
+    calculate_start_end_timesteps(model, positive)
+
+    #make sure each cond area has an opposite one with the same area
+    for c in positive:
+        create_cond_with_same_area_if_none(negative, c)
+    for c in negative:
+        create_cond_with_same_area_if_none(positive, c)
+
+    pre_run_control(model, negative + positive)
+
+    apply_empty_x_to_equal_area(list(filter(lambda c: c.get('control_apply_to_uncond', False) == True, positive)), negative, 'control', lambda cond_cnets, x: cond_cnets[x])
+    apply_empty_x_to_equal_area(positive, negative, 'gligen', lambda cond_cnets, x: cond_cnets[x])
+
+    if latent_image is not None:
+        latent_image = model.process_latent_in(latent_image)
+
+    if hasattr(model, 'extra_conds'):
+        positive = encode_model_conds(model.extra_conds, positive, noise, device, "positive", latent_image=latent_image, denoise_mask=denoise_mask)
+        negative = encode_model_conds(model.extra_conds, negative, noise, device, "negative", latent_image=latent_image, denoise_mask=denoise_mask)
+
+    extra_args = {"cond":positive, "uncond":negative, "cond_scale": cfg, "model_options": model_options, "seed":seed}
+
+    samples = sampler.sample(model_wrap, sigmas, extra_args, callback, noise, latent_image, denoise_mask, disable_pbar)
+    return model.process_latent_out(samples.to(torch.float32))
+
+SCHEDULER_NAMES = ["normal", "karras", "exponential", "sgm_uniform", "simple", "ddim_uniform"]
+SAMPLER_NAMES = KSAMPLER_NAMES + ["ddim", "uni_pc", "uni_pc_bh2"]
+
+def calculate_sigmas_scheduler(model, scheduler_name, steps):
+    if scheduler_name == "karras":
+        sigmas = k_diffusion_sampling.get_sigmas_karras(n=steps, sigma_min=float(model.model_sampling.sigma_min), sigma_max=float(model.model_sampling.sigma_max))
+    elif scheduler_name == "exponential":
+        sigmas = k_diffusion_sampling.get_sigmas_exponential(n=steps, sigma_min=float(model.model_sampling.sigma_min), sigma_max=float(model.model_sampling.sigma_max))
+    elif scheduler_name == "normal":
+        sigmas = normal_scheduler(model, steps)
+    elif scheduler_name == "simple":
+        sigmas = simple_scheduler(model, steps)
+    elif scheduler_name == "ddim_uniform":
+        sigmas = ddim_scheduler(model, steps)
+    elif scheduler_name == "sgm_uniform":
+        sigmas = normal_scheduler(model, steps, sgm=True)
+    else:
+        print("error invalid scheduler", self.scheduler)
+    return sigmas
+
+def sampler_class(name):
+    if name == "uni_pc":
+        sampler = UNIPC
+    elif name == "uni_pc_bh2":
+        sampler = UNIPCBH2
+    elif name == "ddim":
+        sampler = ksampler("euler", inpaint_options={"random": True})
+    else:
+        sampler = ksampler(name)
+    return sampler
+
+class KSampler:
+    SCHEDULERS = SCHEDULER_NAMES
+    SAMPLERS = SAMPLER_NAMES
+
+    def __init__(self, model, steps, device, sampler=None, scheduler=None, denoise=None, model_options={}):
+        self.model = model
+        self.device = device
+        if scheduler not in self.SCHEDULERS:
+            scheduler = self.SCHEDULERS[0]
+        if sampler not in self.SAMPLERS:
+            sampler = self.SAMPLERS[0]
+        self.scheduler = scheduler
+        self.sampler = sampler
+        self.set_steps(steps, denoise)
+        self.denoise = denoise
+        self.model_options = model_options
+
+    def calculate_sigmas(self, steps):
+        sigmas = None
+
+        discard_penultimate_sigma = False
+        if self.sampler in ['dpm_2', 'dpm_2_ancestral', 'uni_pc', 'uni_pc_bh2']:
+            steps += 1
+            discard_penultimate_sigma = True
+
+        sigmas = calculate_sigmas_scheduler(self.model, self.scheduler, steps)
+
+        if discard_penultimate_sigma:
+            sigmas = torch.cat([sigmas[:-2], sigmas[-1:]])
+        return sigmas
+
+    def set_steps(self, steps, denoise=None):
+        self.steps = steps
+        if denoise is None or denoise > 0.9999:
+            self.sigmas = self.calculate_sigmas(steps).to(self.device)
+        else:
+            new_steps = int(steps/denoise)
+            sigmas = self.calculate_sigmas(new_steps).to(self.device)
+            self.sigmas = sigmas[-(steps + 1):]
+
+    def sample(self, noise, positive, negative, cfg, latent_image=None, start_step=None, last_step=None, force_full_denoise=False, denoise_mask=None, sigmas=None, callback=None, disable_pbar=False, seed=None):
+        if sigmas is None:
+            sigmas = self.sigmas
+
+        if last_step is not None and last_step < (len(sigmas) - 1):
+            sigmas = sigmas[:last_step + 1]
+            if force_full_denoise:
+                sigmas[-1] = 0
+
+        if start_step is not None:
+            if start_step < (len(sigmas) - 1):
+                sigmas = sigmas[start_step:]
+            else:
+                if latent_image is not None:
+                    return latent_image
+                else:
+                    return torch.zeros_like(noise)
+
+        sampler = sampler_class(self.sampler)
+
+        return sample(self.model, noise, positive, negative, cfg, self.device, sampler(), sigmas, self.model_options, latent_image=latent_image, denoise_mask=denoise_mask, callback=callback, disable_pbar=disable_pbar, seed=seed)

backend/headless/fcbh/sd.py

@@ -0,0 +1,507 @@
+import torch
+import contextlib
+import math
+
+from fcbh import model_management
+from .ldm.util import instantiate_from_config
+from .ldm.models.autoencoder import AutoencoderKL, AutoencodingEngine
+import yaml
+
+import fcbh.utils
+
+from . import clip_vision
+from . import gligen
+from . import diffusers_convert
+from . import model_base
+from . import model_detection
+
+from . import sd1_clip
+from . import sd2_clip
+from . import sdxl_clip
+
+import fcbh.model_patcher
+import fcbh.lora
+import fcbh.t2i_adapter.adapter
+import fcbh.supported_models_base
+
+def load_model_weights(model, sd):
+    m, u = model.load_state_dict(sd, strict=False)
+    m = set(m)
+    unexpected_keys = set(u)
+
+    k = list(sd.keys())
+    for x in k:
+        if x not in unexpected_keys:
+            w = sd.pop(x)
+            del w
+    if len(m) > 0:
+        print("missing", m)
+    return model
+
+def load_clip_weights(model, sd):
+    k = list(sd.keys())
+    for x in k:
+        if x.startswith("cond_stage_model.transformer.") and not x.startswith("cond_stage_model.transformer.text_model."):
+            y = x.replace("cond_stage_model.transformer.", "cond_stage_model.transformer.text_model.")
+            sd[y] = sd.pop(x)
+
+    if 'cond_stage_model.transformer.text_model.embeddings.position_ids' in sd:
+        ids = sd['cond_stage_model.transformer.text_model.embeddings.position_ids']
+        if ids.dtype == torch.float32:
+            sd['cond_stage_model.transformer.text_model.embeddings.position_ids'] = ids.round()
+
+    sd = fcbh.utils.transformers_convert(sd, "cond_stage_model.model.", "cond_stage_model.transformer.text_model.", 24)
+    return load_model_weights(model, sd)
+
+
+def load_lora_for_models(model, clip, lora, strength_model, strength_clip):
+    key_map = {}
+    if model is not None:
+        key_map = fcbh.lora.model_lora_keys_unet(model.model, key_map)
+    if clip is not None:
+        key_map = fcbh.lora.model_lora_keys_clip(clip.cond_stage_model, key_map)
+
+    loaded = fcbh.lora.load_lora(lora, key_map)
+    if model is not None:
+        new_modelpatcher = model.clone()
+        k = new_modelpatcher.add_patches(loaded, strength_model)
+    else:
+        k = ()
+        new_modelpatcher = None
+
+    if clip is not None:
+        new_clip = clip.clone()
+        k1 = new_clip.add_patches(loaded, strength_clip)
+    else:
+        k1 = ()
+        new_clip = None
+    k = set(k)
+    k1 = set(k1)
+    for x in loaded:
+        if (x not in k) and (x not in k1):
+            print("NOT LOADED", x)
+
+    return (new_modelpatcher, new_clip)
+
+
+class CLIP:
+    def __init__(self, target=None, embedding_directory=None, no_init=False):
+        if no_init:
+            return
+        params = target.params.copy()
+        clip = target.clip
+        tokenizer = target.tokenizer
+
+        load_device = model_management.text_encoder_device()
+        offload_device = model_management.text_encoder_offload_device()
+        params['device'] = offload_device
+        if model_management.should_use_fp16(load_device, prioritize_performance=False):
+            params['dtype'] = torch.float16
+        else:
+            params['dtype'] = torch.float32
+
+        self.cond_stage_model = clip(**(params))
+
+        self.tokenizer = tokenizer(embedding_directory=embedding_directory)
+        self.patcher = fcbh.model_patcher.ModelPatcher(self.cond_stage_model, load_device=load_device, offload_device=offload_device)
+        self.layer_idx = None
+
+    def clone(self):
+        n = CLIP(no_init=True)
+        n.patcher = self.patcher.clone()
+        n.cond_stage_model = self.cond_stage_model
+        n.tokenizer = self.tokenizer
+        n.layer_idx = self.layer_idx
+        return n
+
+    def add_patches(self, patches, strength_patch=1.0, strength_model=1.0):
+        return self.patcher.add_patches(patches, strength_patch, strength_model)
+
+    def clip_layer(self, layer_idx):
+        self.layer_idx = layer_idx
+
+    def tokenize(self, text, return_word_ids=False):
+        return self.tokenizer.tokenize_with_weights(text, return_word_ids)
+
+    def encode_from_tokens(self, tokens, return_pooled=False):
+        if self.layer_idx is not None:
+            self.cond_stage_model.clip_layer(self.layer_idx)
+        else:
+            self.cond_stage_model.reset_clip_layer()
+
+        self.load_model()
+        cond, pooled = self.cond_stage_model.encode_token_weights(tokens)
+        if return_pooled:
+            return cond, pooled
+        return cond
+
+    def encode(self, text):
+        tokens = self.tokenize(text)
+        return self.encode_from_tokens(tokens)
+
+    def load_sd(self, sd):
+        return self.cond_stage_model.load_sd(sd)
+
+    def get_sd(self):
+        return self.cond_stage_model.state_dict()
+
+    def load_model(self):
+        model_management.load_model_gpu(self.patcher)
+        return self.patcher
+
+    def get_key_patches(self):
+        return self.patcher.get_key_patches()
+
+class VAE:
+    def __init__(self, sd=None, device=None, config=None):
+        if 'decoder.up_blocks.0.resnets.0.norm1.weight' in sd.keys(): #diffusers format
+            sd = diffusers_convert.convert_vae_state_dict(sd)
+
+        if config is None:
+            #default SD1.x/SD2.x VAE parameters
+            ddconfig = {'double_z': True, 'z_channels': 4, 'resolution': 256, 'in_channels': 3, 'out_ch': 3, 'ch': 128, 'ch_mult': [1, 2, 4, 4], 'num_res_blocks': 2, 'attn_resolutions': [], 'dropout': 0.0}
+            self.first_stage_model = AutoencoderKL(ddconfig=ddconfig, embed_dim=4)
+        else:
+            self.first_stage_model = AutoencoderKL(**(config['params']))
+        self.first_stage_model = self.first_stage_model.eval()
+
+        m, u = self.first_stage_model.load_state_dict(sd, strict=False)
+        if len(m) > 0:
+            print("Missing VAE keys", m)
+
+        if len(u) > 0:
+            print("Leftover VAE keys", u)
+
+        if device is None:
+            device = model_management.vae_device()
+        self.device = device
+        self.offload_device = model_management.vae_offload_device()
+        self.vae_dtype = model_management.vae_dtype()
+        self.first_stage_model.to(self.vae_dtype)
+
+    def decode_tiled_(self, samples, tile_x=64, tile_y=64, overlap = 16):
+        steps = samples.shape[0] * fcbh.utils.get_tiled_scale_steps(samples.shape[3], samples.shape[2], tile_x, tile_y, overlap)
+        steps += samples.shape[0] * fcbh.utils.get_tiled_scale_steps(samples.shape[3], samples.shape[2], tile_x // 2, tile_y * 2, overlap)
+        steps += samples.shape[0] * fcbh.utils.get_tiled_scale_steps(samples.shape[3], samples.shape[2], tile_x * 2, tile_y // 2, overlap)
+        pbar = fcbh.utils.ProgressBar(steps)
+
+        decode_fn = lambda a: (self.first_stage_model.decode(a.to(self.vae_dtype).to(self.device)) + 1.0).float()
+        output = torch.clamp((
+            (fcbh.utils.tiled_scale(samples, decode_fn, tile_x // 2, tile_y * 2, overlap, upscale_amount = 8, pbar = pbar) +
+            fcbh.utils.tiled_scale(samples, decode_fn, tile_x * 2, tile_y // 2, overlap, upscale_amount = 8, pbar = pbar) +
+             fcbh.utils.tiled_scale(samples, decode_fn, tile_x, tile_y, overlap, upscale_amount = 8, pbar = pbar))
+            / 3.0) / 2.0, min=0.0, max=1.0)
+        return output
+
+    def encode_tiled_(self, pixel_samples, tile_x=512, tile_y=512, overlap = 64):
+        steps = pixel_samples.shape[0] * fcbh.utils.get_tiled_scale_steps(pixel_samples.shape[3], pixel_samples.shape[2], tile_x, tile_y, overlap)
+        steps += pixel_samples.shape[0] * fcbh.utils.get_tiled_scale_steps(pixel_samples.shape[3], pixel_samples.shape[2], tile_x // 2, tile_y * 2, overlap)
+        steps += pixel_samples.shape[0] * fcbh.utils.get_tiled_scale_steps(pixel_samples.shape[3], pixel_samples.shape[2], tile_x * 2, tile_y // 2, overlap)
+        pbar = fcbh.utils.ProgressBar(steps)
+
+        encode_fn = lambda a: self.first_stage_model.encode((2. * a - 1.).to(self.vae_dtype).to(self.device)).float()
+        samples = fcbh.utils.tiled_scale(pixel_samples, encode_fn, tile_x, tile_y, overlap, upscale_amount = (1/8), out_channels=4, pbar=pbar)
+        samples += fcbh.utils.tiled_scale(pixel_samples, encode_fn, tile_x * 2, tile_y // 2, overlap, upscale_amount = (1/8), out_channels=4, pbar=pbar)
+        samples += fcbh.utils.tiled_scale(pixel_samples, encode_fn, tile_x // 2, tile_y * 2, overlap, upscale_amount = (1/8), out_channels=4, pbar=pbar)
+        samples /= 3.0
+        return samples
+
+    def decode(self, samples_in):
+        self.first_stage_model = self.first_stage_model.to(self.device)
+        try:
+            memory_used = (2562 * samples_in.shape[2] * samples_in.shape[3] * 64) * 1.7
+            model_management.free_memory(memory_used, self.device)
+            free_memory = model_management.get_free_memory(self.device)
+            batch_number = int(free_memory / memory_used)
+            batch_number = max(1, batch_number)
+
+            pixel_samples = torch.empty((samples_in.shape[0], 3, round(samples_in.shape[2] * 8), round(samples_in.shape[3] * 8)), device="cpu")
+            for x in range(0, samples_in.shape[0], batch_number):
+                samples = samples_in[x:x+batch_number].to(self.vae_dtype).to(self.device)
+                pixel_samples[x:x+batch_number] = torch.clamp((self.first_stage_model.decode(samples).cpu().float() + 1.0) / 2.0, min=0.0, max=1.0)
+        except model_management.OOM_EXCEPTION as e:
+            print("Warning: Ran out of memory when regular VAE decoding, retrying with tiled VAE decoding.")
+            pixel_samples = self.decode_tiled_(samples_in)
+
+        self.first_stage_model = self.first_stage_model.to(self.offload_device)
+        pixel_samples = pixel_samples.cpu().movedim(1,-1)
+        return pixel_samples
+
+    def decode_tiled(self, samples, tile_x=64, tile_y=64, overlap = 16):
+        self.first_stage_model = self.first_stage_model.to(self.device)
+        output = self.decode_tiled_(samples, tile_x, tile_y, overlap)
+        self.first_stage_model = self.first_stage_model.to(self.offload_device)
+        return output.movedim(1,-1)
+
+    def encode(self, pixel_samples):
+        self.first_stage_model = self.first_stage_model.to(self.device)
+        pixel_samples = pixel_samples.movedim(-1,1)
+        try:
+            memory_used = (2078 * pixel_samples.shape[2] * pixel_samples.shape[3]) * 1.7 #NOTE: this constant along with the one in the decode above are estimated from the mem usage for the VAE and could change.
+            model_management.free_memory(memory_used, self.device)
+            free_memory = model_management.get_free_memory(self.device)
+            batch_number = int(free_memory / memory_used)
+            batch_number = max(1, batch_number)
+            samples = torch.empty((pixel_samples.shape[0], 4, round(pixel_samples.shape[2] // 8), round(pixel_samples.shape[3] // 8)), device="cpu")
+            for x in range(0, pixel_samples.shape[0], batch_number):
+                pixels_in = (2. * pixel_samples[x:x+batch_number] - 1.).to(self.vae_dtype).to(self.device)
+                samples[x:x+batch_number] = self.first_stage_model.encode(pixels_in).cpu().float()
+
+        except model_management.OOM_EXCEPTION as e:
+            print("Warning: Ran out of memory when regular VAE encoding, retrying with tiled VAE encoding.")
+            samples = self.encode_tiled_(pixel_samples)
+
+        self.first_stage_model = self.first_stage_model.to(self.offload_device)
+        return samples
+
+    def encode_tiled(self, pixel_samples, tile_x=512, tile_y=512, overlap = 64):
+        self.first_stage_model = self.first_stage_model.to(self.device)
+        pixel_samples = pixel_samples.movedim(-1,1)
+        samples = self.encode_tiled_(pixel_samples, tile_x=tile_x, tile_y=tile_y, overlap=overlap)
+        self.first_stage_model = self.first_stage_model.to(self.offload_device)
+        return samples
+
+    def get_sd(self):
+        return self.first_stage_model.state_dict()
+
+class StyleModel:
+    def __init__(self, model, device="cpu"):
+        self.model = model
+
+    def get_cond(self, input):
+        return self.model(input.last_hidden_state)
+
+
+def load_style_model(ckpt_path):
+    model_data = fcbh.utils.load_torch_file(ckpt_path, safe_load=True)
+    keys = model_data.keys()
+    if "style_embedding" in keys:
+        model = fcbh.t2i_adapter.adapter.StyleAdapter(width=1024, context_dim=768, num_head=8, n_layes=3, num_token=8)
+    else:
+        raise Exception("invalid style model {}".format(ckpt_path))
+    model.load_state_dict(model_data)
+    return StyleModel(model)
+
+
+def load_clip(ckpt_paths, embedding_directory=None):
+    clip_data = []
+    for p in ckpt_paths:
+        clip_data.append(fcbh.utils.load_torch_file(p, safe_load=True))
+
+    class EmptyClass:
+        pass
+
+    for i in range(len(clip_data)):
+        if "transformer.resblocks.0.ln_1.weight" in clip_data[i]:
+            clip_data[i] = fcbh.utils.transformers_convert(clip_data[i], "", "text_model.", 32)
+
+    clip_target = EmptyClass()
+    clip_target.params = {}
+    if len(clip_data) == 1:
+        if "text_model.encoder.layers.30.mlp.fc1.weight" in clip_data[0]:
+            clip_target.clip = sdxl_clip.SDXLRefinerClipModel
+            clip_target.tokenizer = sdxl_clip.SDXLTokenizer
+        elif "text_model.encoder.layers.22.mlp.fc1.weight" in clip_data[0]:
+            clip_target.clip = sd2_clip.SD2ClipModel
+            clip_target.tokenizer = sd2_clip.SD2Tokenizer
+        else:
+            clip_target.clip = sd1_clip.SD1ClipModel
+            clip_target.tokenizer = sd1_clip.SD1Tokenizer
+    else:
+        clip_target.clip = sdxl_clip.SDXLClipModel
+        clip_target.tokenizer = sdxl_clip.SDXLTokenizer
+
+    clip = CLIP(clip_target, embedding_directory=embedding_directory)
+    for c in clip_data:
+        m, u = clip.load_sd(c)
+        if len(m) > 0:
+            print("clip missing:", m)
+
+        if len(u) > 0:
+            print("clip unexpected:", u)
+    return clip
+
+def load_gligen(ckpt_path):
+    data = fcbh.utils.load_torch_file(ckpt_path, safe_load=True)
+    model = gligen.load_gligen(data)
+    if model_management.should_use_fp16():
+        model = model.half()
+    return fcbh.model_patcher.ModelPatcher(model, load_device=model_management.get_torch_device(), offload_device=model_management.unet_offload_device())
+
+def load_checkpoint(config_path=None, ckpt_path=None, output_vae=True, output_clip=True, embedding_directory=None, state_dict=None, config=None):
+    #TODO: this function is a mess and should be removed eventually
+    if config is None:
+        with open(config_path, 'r') as stream:
+            config = yaml.safe_load(stream)
+    model_config_params = config['model']['params']
+    clip_config = model_config_params['cond_stage_config']
+    scale_factor = model_config_params['scale_factor']
+    vae_config = model_config_params['first_stage_config']
+
+    fp16 = False
+    if "unet_config" in model_config_params:
+        if "params" in model_config_params["unet_config"]:
+            unet_config = model_config_params["unet_config"]["params"]
+            if "use_fp16" in unet_config:
+                fp16 = unet_config.pop("use_fp16")
+                if fp16:
+                    unet_config["dtype"] = torch.float16
+
+    noise_aug_config = None
+    if "noise_aug_config" in model_config_params:
+        noise_aug_config = model_config_params["noise_aug_config"]
+
+    model_type = model_base.ModelType.EPS
+
+    if "parameterization" in model_config_params:
+        if model_config_params["parameterization"] == "v":
+            model_type = model_base.ModelType.V_PREDICTION
+
+    clip = None
+    vae = None
+
+    class WeightsLoader(torch.nn.Module):
+        pass
+
+    if state_dict is None:
+        state_dict = fcbh.utils.load_torch_file(ckpt_path)
+
+    class EmptyClass:
+        pass
+
+    model_config = fcbh.supported_models_base.BASE({})
+
+    from . import latent_formats
+    model_config.latent_format = latent_formats.SD15(scale_factor=scale_factor)
+    model_config.unet_config = model_detection.convert_config(unet_config)
+
+    if config['model']["target"].endswith("ImageEmbeddingConditionedLatentDiffusion"):
+        model = model_base.SD21UNCLIP(model_config, noise_aug_config["params"], model_type=model_type)
+    else:
+        model = model_base.BaseModel(model_config, model_type=model_type)
+
+    if config['model']["target"].endswith("LatentInpaintDiffusion"):
+        model.set_inpaint()
+
+    if fp16:
+        model = model.half()
+
+    offload_device = model_management.unet_offload_device()
+    model = model.to(offload_device)
+    model.load_model_weights(state_dict, "model.diffusion_model.")
+
+    if output_vae:
+        vae_sd = fcbh.utils.state_dict_prefix_replace(state_dict, {"first_stage_model.": ""}, filter_keys=True)
+        vae = VAE(sd=vae_sd, config=vae_config)
+
+    if output_clip:
+        w = WeightsLoader()
+        clip_target = EmptyClass()
+        clip_target.params = clip_config.get("params", {})
+        if clip_config["target"].endswith("FrozenOpenCLIPEmbedder"):
+            clip_target.clip = sd2_clip.SD2ClipModel
+            clip_target.tokenizer = sd2_clip.SD2Tokenizer
+            clip = CLIP(clip_target, embedding_directory=embedding_directory)
+            w.cond_stage_model = clip.cond_stage_model.clip_h
+        elif clip_config["target"].endswith("FrozenCLIPEmbedder"):
+            clip_target.clip = sd1_clip.SD1ClipModel
+            clip_target.tokenizer = sd1_clip.SD1Tokenizer
+            clip = CLIP(clip_target, embedding_directory=embedding_directory)
+            w.cond_stage_model = clip.cond_stage_model.clip_l
+        load_clip_weights(w, state_dict)
+
+    return (fcbh.model_patcher.ModelPatcher(model, load_device=model_management.get_torch_device(), offload_device=offload_device), clip, vae)
+
+def load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, output_clipvision=False, embedding_directory=None, output_model=True):
+    sd = fcbh.utils.load_torch_file(ckpt_path)
+    sd_keys = sd.keys()
+    clip = None
+    clipvision = None
+    vae = None
+    model = None
+    model_patcher = None
+    clip_target = None
+
+    parameters = fcbh.utils.calculate_parameters(sd, "model.diffusion_model.")
+    unet_dtype = model_management.unet_dtype(model_params=parameters)
+
+    class WeightsLoader(torch.nn.Module):
+        pass
+
+    model_config = model_detection.model_config_from_unet(sd, "model.diffusion_model.", unet_dtype)
+    if model_config is None:
+        raise RuntimeError("ERROR: Could not detect model type of: {}".format(ckpt_path))
+
+    if model_config.clip_vision_prefix is not None:
+        if output_clipvision:
+            clipvision = clip_vision.load_clipvision_from_sd(sd, model_config.clip_vision_prefix, True)
+
+    if output_model:
+        inital_load_device = model_management.unet_inital_load_device(parameters, unet_dtype)
+        offload_device = model_management.unet_offload_device()
+        model = model_config.get_model(sd, "model.diffusion_model.", device=inital_load_device)
+        model.load_model_weights(sd, "model.diffusion_model.")
+
+    if output_vae:
+        vae_sd = fcbh.utils.state_dict_prefix_replace(sd, {"first_stage_model.": ""}, filter_keys=True)
+        vae = VAE(sd=vae_sd)
+
+    if output_clip:
+        w = WeightsLoader()
+        clip_target = model_config.clip_target()
+        if clip_target is not None:
+            clip = CLIP(clip_target, embedding_directory=embedding_directory)
+            w.cond_stage_model = clip.cond_stage_model
+            sd = model_config.process_clip_state_dict(sd)
+            load_model_weights(w, sd)
+
+    left_over = sd.keys()
+    if len(left_over) > 0:
+        print("left over keys:", left_over)
+
+    if output_model:
+        model_patcher = fcbh.model_patcher.ModelPatcher(model, load_device=model_management.get_torch_device(), offload_device=model_management.unet_offload_device(), current_device=inital_load_device)
+        if inital_load_device != torch.device("cpu"):
+            print("loaded straight to GPU")
+            model_management.load_model_gpu(model_patcher)
+
+    return (model_patcher, clip, vae, clipvision)
+
+
+def load_unet(unet_path): #load unet in diffusers format
+    sd = fcbh.utils.load_torch_file(unet_path)
+    parameters = fcbh.utils.calculate_parameters(sd)
+    unet_dtype = model_management.unet_dtype(model_params=parameters)
+    if "input_blocks.0.0.weight" in sd: #ldm
+        model_config = model_detection.model_config_from_unet(sd, "", unet_dtype)
+        if model_config is None:
+            raise RuntimeError("ERROR: Could not detect model type of: {}".format(unet_path))
+        new_sd = sd
+
+    else: #diffusers
+        model_config = model_detection.model_config_from_diffusers_unet(sd, unet_dtype)
+        if model_config is None:
+            print("ERROR UNSUPPORTED UNET", unet_path)
+            return None
+
+        diffusers_keys = fcbh.utils.unet_to_diffusers(model_config.unet_config)
+
+        new_sd = {}
+        for k in diffusers_keys:
+            if k in sd:
+                new_sd[diffusers_keys[k]] = sd.pop(k)
+            else:
+                print(diffusers_keys[k], k)
+    offload_device = model_management.unet_offload_device()
+    model = model_config.get_model(new_sd, "")
+    model = model.to(offload_device)
+    model.load_model_weights(new_sd, "")
+    left_over = sd.keys()
+    if len(left_over) > 0:
+        print("left over keys in unet:", left_over)
+    return fcbh.model_patcher.ModelPatcher(model, load_device=model_management.get_torch_device(), offload_device=offload_device)
+
+def save_checkpoint(output_path, model, clip, vae, metadata=None):
+    model_management.load_models_gpu([model, clip.load_model()])
+    sd = model.model.state_dict_for_saving(clip.get_sd(), vae.get_sd())
+    fcbh.utils.save_torch_file(sd, output_path, metadata=metadata)

backend/headless/fcbh/sd1_clip.py

@@ -0,0 +1,539 @@
+import os
+
+from transformers import CLIPTokenizer, CLIPTextModel, CLIPTextConfig, modeling_utils
+import fcbh.ops
+import torch
+import traceback
+import zipfile
+from . import model_management
+import contextlib
+
+def gen_empty_tokens(special_tokens, length):
+    start_token = special_tokens.get("start", None)
+    end_token = special_tokens.get("end", None)
+    pad_token = special_tokens.get("pad")
+    output = []
+    if start_token is not None:
+        output.append(start_token)
+    if end_token is not None:
+        output.append(end_token)
+    output += [pad_token] * (length - len(output))
+    return output
+
+class ClipTokenWeightEncoder:
+    def encode_token_weights(self, token_weight_pairs):
+        to_encode = list()
+        max_token_len = 0
+        has_weights = False
+        for x in token_weight_pairs:
+            tokens = list(map(lambda a: a[0], x))
+            max_token_len = max(len(tokens), max_token_len)
+            has_weights = has_weights or not all(map(lambda a: a[1] == 1.0, x))
+            to_encode.append(tokens)
+
+        sections = len(to_encode)
+        if has_weights or sections == 0:
+            to_encode.append(gen_empty_tokens(self.special_tokens, max_token_len))
+
+        out, pooled = self.encode(to_encode)
+        if pooled is not None:
+            first_pooled = pooled[0:1].cpu()
+        else:
+            first_pooled = pooled
+
+        output = []
+        for k in range(0, sections):
+            z = out[k:k+1]
+            if has_weights:
+                z_empty = out[-1]
+                for i in range(len(z)):
+                    for j in range(len(z[i])):
+                        weight = token_weight_pairs[k][j][1]
+                        if weight != 1.0:
+                            z[i][j] = (z[i][j] - z_empty[j]) * weight + z_empty[j]
+            output.append(z)
+
+        if (len(output) == 0):
+            return out[-1:].cpu(), first_pooled
+        return torch.cat(output, dim=-2).cpu(), first_pooled
+
+class SDClipModel(torch.nn.Module, ClipTokenWeightEncoder):
+    """Uses the CLIP transformer encoder for text (from huggingface)"""
+    LAYERS = [
+        "last",
+        "pooled",
+        "hidden"
+    ]
+    def __init__(self, version="openai/clip-vit-large-patch14", device="cpu", max_length=77,
+                 freeze=True, layer="last", layer_idx=None, textmodel_json_config=None, textmodel_path=None, dtype=None,
+                 special_tokens={"start": 49406, "end": 49407, "pad": 49407},layer_norm_hidden_state=True, config_class=CLIPTextConfig,
+                 model_class=CLIPTextModel, inner_name="text_model"):  # clip-vit-base-patch32
+        super().__init__()
+        assert layer in self.LAYERS
+        self.num_layers = 12
+        if textmodel_path is not None:
+            self.transformer = model_class.from_pretrained(textmodel_path)
+        else:
+            if textmodel_json_config is None:
+                textmodel_json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "sd1_clip_config.json")
+            config = config_class.from_json_file(textmodel_json_config)
+            self.num_layers = config.num_hidden_layers
+            with fcbh.ops.use_fcbh_ops(device, dtype):
+                with modeling_utils.no_init_weights():
+                    self.transformer = model_class(config)
+
+        self.inner_name = inner_name
+        if dtype is not None:
+            self.transformer.to(dtype)
+            inner_model = getattr(self.transformer, self.inner_name)
+            if hasattr(inner_model, "embeddings"):
+                inner_model.embeddings.to(torch.float32)
+            else:
+                self.transformer.set_input_embeddings(self.transformer.get_input_embeddings().to(torch.float32))
+
+        self.max_length = max_length
+        if freeze:
+            self.freeze()
+        self.layer = layer
+        self.layer_idx = None
+        self.special_tokens = special_tokens
+        self.text_projection = torch.nn.Parameter(torch.eye(self.transformer.get_input_embeddings().weight.shape[1]))
+        self.logit_scale = torch.nn.Parameter(torch.tensor(4.6055))
+        self.enable_attention_masks = False
+
+        self.layer_norm_hidden_state = layer_norm_hidden_state
+        if layer == "hidden":
+            assert layer_idx is not None
+            assert abs(layer_idx) <= self.num_layers
+            self.clip_layer(layer_idx)
+        self.layer_default = (self.layer, self.layer_idx)
+
+    def freeze(self):
+        self.transformer = self.transformer.eval()
+        #self.train = disabled_train
+        for param in self.parameters():
+            param.requires_grad = False
+
+    def clip_layer(self, layer_idx):
+        if abs(layer_idx) >= self.num_layers:
+            self.layer = "last"
+        else:
+            self.layer = "hidden"
+            self.layer_idx = layer_idx
+
+    def reset_clip_layer(self):
+        self.layer = self.layer_default[0]
+        self.layer_idx = self.layer_default[1]
+
+    def set_up_textual_embeddings(self, tokens, current_embeds):
+        out_tokens = []
+        next_new_token = token_dict_size = current_embeds.weight.shape[0] - 1
+        embedding_weights = []
+
+        for x in tokens:
+            tokens_temp = []
+            for y in x:
+                if isinstance(y, int):
+                    if y == token_dict_size: #EOS token
+                        y = -1
+                    tokens_temp += [y]
+                else:
+                    if y.shape[0] == current_embeds.weight.shape[1]:
+                        embedding_weights += [y]
+                        tokens_temp += [next_new_token]
+                        next_new_token += 1
+                    else:
+                        print("WARNING: shape mismatch when trying to apply embedding, embedding will be ignored", y.shape[0], current_embeds.weight.shape[1])
+            while len(tokens_temp) < len(x):
+                tokens_temp += [self.special_tokens["pad"]]
+            out_tokens += [tokens_temp]
+
+        n = token_dict_size
+        if len(embedding_weights) > 0:
+            new_embedding = torch.nn.Embedding(next_new_token + 1, current_embeds.weight.shape[1], device=current_embeds.weight.device, dtype=current_embeds.weight.dtype)
+            new_embedding.weight[:token_dict_size] = current_embeds.weight[:-1]
+            for x in embedding_weights:
+                new_embedding.weight[n] = x
+                n += 1
+            new_embedding.weight[n] = current_embeds.weight[-1] #EOS embedding
+            self.transformer.set_input_embeddings(new_embedding)
+
+        processed_tokens = []
+        for x in out_tokens:
+            processed_tokens += [list(map(lambda a: n if a == -1 else a, x))] #The EOS token should always be the largest one
+
+        return processed_tokens
+
+    def forward(self, tokens):
+        backup_embeds = self.transformer.get_input_embeddings()
+        device = backup_embeds.weight.device
+        tokens = self.set_up_textual_embeddings(tokens, backup_embeds)
+        tokens = torch.LongTensor(tokens).to(device)
+
+        if getattr(self.transformer, self.inner_name).final_layer_norm.weight.dtype != torch.float32:
+            precision_scope = torch.autocast
+        else:
+            precision_scope = lambda a, b: contextlib.nullcontext(a)
+
+        with precision_scope(model_management.get_autocast_device(device), torch.float32):
+            attention_mask = None
+            if self.enable_attention_masks:
+                attention_mask = torch.zeros_like(tokens)
+                max_token = self.transformer.get_input_embeddings().weight.shape[0] - 1
+                for x in range(attention_mask.shape[0]):
+                    for y in range(attention_mask.shape[1]):
+                        attention_mask[x, y] = 1
+                        if tokens[x, y] == max_token:
+                            break
+
+            outputs = self.transformer(input_ids=tokens, attention_mask=attention_mask, output_hidden_states=self.layer=="hidden")
+            self.transformer.set_input_embeddings(backup_embeds)
+
+            if self.layer == "last":
+                z = outputs.last_hidden_state
+            elif self.layer == "pooled":
+                z = outputs.pooler_output[:, None, :]
+            else:
+                z = outputs.hidden_states[self.layer_idx]
+                if self.layer_norm_hidden_state:
+                    z = getattr(self.transformer, self.inner_name).final_layer_norm(z)
+
+            if hasattr(outputs, "pooler_output"):
+                pooled_output = outputs.pooler_output.float()
+            else:
+                pooled_output = None
+
+            if self.text_projection is not None and pooled_output is not None:
+                pooled_output = pooled_output.float().to(self.text_projection.device) @ self.text_projection.float()
+        return z.float(), pooled_output
+
+    def encode(self, tokens):
+        return self(tokens)
+
+    def load_sd(self, sd):
+        if "text_projection" in sd:
+            self.text_projection[:] = sd.pop("text_projection")
+        if "text_projection.weight" in sd:
+            self.text_projection[:] = sd.pop("text_projection.weight").transpose(0, 1)
+        return self.transformer.load_state_dict(sd, strict=False)
+
+def parse_parentheses(string):
+    result = []
+    current_item = ""
+    nesting_level = 0
+    for char in string:
+        if char == "(":
+            if nesting_level == 0:
+                if current_item:
+                    result.append(current_item)
+                    current_item = "("
+                else:
+                    current_item = "("
+            else:
+                current_item += char
+            nesting_level += 1
+        elif char == ")":
+            nesting_level -= 1
+            if nesting_level == 0:
+                result.append(current_item + ")")
+                current_item = ""
+            else:
+                current_item += char
+        else:
+            current_item += char
+    if current_item:
+        result.append(current_item)
+    return result
+
+def token_weights(string, current_weight):
+    a = parse_parentheses(string)
+    out = []
+    for x in a:
+        weight = current_weight
+        if len(x) >= 2 and x[-1] == ')' and x[0] == '(':
+            x = x[1:-1]
+            xx = x.rfind(":")
+            weight *= 1.1
+            if xx > 0:
+                try:
+                    weight = float(x[xx+1:])
+                    x = x[:xx]
+                except:
+                    pass
+            out += token_weights(x, weight)
+        else:
+            out += [(x, current_weight)]
+    return out
+
+def escape_important(text):
+    text = text.replace("\\)", "\0\1")
+    text = text.replace("\\(", "\0\2")
+    return text
+
+def unescape_important(text):
+    text = text.replace("\0\1", ")")
+    text = text.replace("\0\2", "(")
+    return text
+
+def safe_load_embed_zip(embed_path):
+    with zipfile.ZipFile(embed_path) as myzip:
+        names = list(filter(lambda a: "data/" in a, myzip.namelist()))
+        names.reverse()
+        for n in names:
+            with myzip.open(n) as myfile:
+                data = myfile.read()
+                number = len(data) // 4
+                length_embed = 1024 #sd2.x
+                if number < 768:
+                    continue
+                if number % 768 == 0:
+                    length_embed = 768 #sd1.x
+                num_embeds = number // length_embed
+                embed = torch.frombuffer(data, dtype=torch.float)
+                out = embed.reshape((num_embeds, length_embed)).clone()
+                del embed
+                return out
+
+def expand_directory_list(directories):
+    dirs = set()
+    for x in directories:
+        dirs.add(x)
+        for root, subdir, file in os.walk(x, followlinks=True):
+            dirs.add(root)
+    return list(dirs)
+
+def load_embed(embedding_name, embedding_directory, embedding_size, embed_key=None):
+    if isinstance(embedding_directory, str):
+        embedding_directory = [embedding_directory]
+
+    embedding_directory = expand_directory_list(embedding_directory)
+
+    valid_file = None
+    for embed_dir in embedding_directory:
+        embed_path = os.path.abspath(os.path.join(embed_dir, embedding_name))
+        embed_dir = os.path.abspath(embed_dir)
+        try:
+            if os.path.commonpath((embed_dir, embed_path)) != embed_dir:
+                continue
+        except:
+            continue
+        if not os.path.isfile(embed_path):
+            extensions = ['.safetensors', '.pt', '.bin']
+            for x in extensions:
+                t = embed_path + x
+                if os.path.isfile(t):
+                    valid_file = t
+                    break
+        else:
+            valid_file = embed_path
+        if valid_file is not None:
+            break
+
+    if valid_file is None:
+        return None
+
+    embed_path = valid_file
+
+    embed_out = None
+
+    try:
+        if embed_path.lower().endswith(".safetensors"):
+            import safetensors.torch
+            embed = safetensors.torch.load_file(embed_path, device="cpu")
+        else:
+            if 'weights_only' in torch.load.__code__.co_varnames:
+                try:
+                    embed = torch.load(embed_path, weights_only=True, map_location="cpu")
+                except:
+                    embed_out = safe_load_embed_zip(embed_path)
+            else:
+                embed = torch.load(embed_path, map_location="cpu")
+    except Exception as e:
+        print(traceback.format_exc())
+        print()
+        print("error loading embedding, skipping loading:", embedding_name)
+        return None
+
+    if embed_out is None:
+        if 'string_to_param' in embed:
+            values = embed['string_to_param'].values()
+            embed_out = next(iter(values))
+        elif isinstance(embed, list):
+            out_list = []
+            for x in range(len(embed)):
+                for k in embed[x]:
+                    t = embed[x][k]
+                    if t.shape[-1] != embedding_size:
+                        continue
+                    out_list.append(t.reshape(-1, t.shape[-1]))
+            embed_out = torch.cat(out_list, dim=0)
+        elif embed_key is not None and embed_key in embed:
+            embed_out = embed[embed_key]
+        else:
+            values = embed.values()
+            embed_out = next(iter(values))
+    return embed_out
+
+class SDTokenizer:
+    def __init__(self, tokenizer_path=None, max_length=77, pad_with_end=True, embedding_directory=None, embedding_size=768, embedding_key='clip_l', tokenizer_class=CLIPTokenizer, has_start_token=True, pad_to_max_length=True):
+        if tokenizer_path is None:
+            tokenizer_path = os.path.join(os.path.dirname(os.path.realpath(__file__)), "sd1_tokenizer")
+        self.tokenizer = tokenizer_class.from_pretrained(tokenizer_path)
+        self.max_length = max_length
+
+        empty = self.tokenizer('')["input_ids"]
+        if has_start_token:
+            self.tokens_start = 1
+            self.start_token = empty[0]
+            self.end_token = empty[1]
+        else:
+            self.tokens_start = 0
+            self.start_token = None
+            self.end_token = empty[0]
+        self.pad_with_end = pad_with_end
+        self.pad_to_max_length = pad_to_max_length
+
+        vocab = self.tokenizer.get_vocab()
+        self.inv_vocab = {v: k for k, v in vocab.items()}
+        self.embedding_directory = embedding_directory
+        self.max_word_length = 8
+        self.embedding_identifier = "embedding:"
+        self.embedding_size = embedding_size
+        self.embedding_key = embedding_key
+
+    def _try_get_embedding(self, embedding_name:str):
+        '''
+        Takes a potential embedding name and tries to retrieve it.
+        Returns a Tuple consisting of the embedding and any leftover string, embedding can be None.
+        '''
+        embed = load_embed(embedding_name, self.embedding_directory, self.embedding_size, self.embedding_key)
+        if embed is None:
+            stripped = embedding_name.strip(',')
+            if len(stripped) < len(embedding_name):
+                embed = load_embed(stripped, self.embedding_directory, self.embedding_size, self.embedding_key)
+                return (embed, embedding_name[len(stripped):])
+        return (embed, "")
+
+
+    def tokenize_with_weights(self, text:str, return_word_ids=False):
+        '''
+        Takes a prompt and converts it to a list of (token, weight, word id) elements.
+        Tokens can both be integer tokens and pre computed CLIP tensors.
+        Word id values are unique per word and embedding, where the id 0 is reserved for non word tokens.
+        Returned list has the dimensions NxM where M is the input size of CLIP
+        '''
+        if self.pad_with_end:
+            pad_token = self.end_token
+        else:
+            pad_token = 0
+
+        text = escape_important(text)
+        parsed_weights = token_weights(text, 1.0)
+
+        #tokenize words
+        tokens = []
+        for weighted_segment, weight in parsed_weights:
+            to_tokenize = unescape_important(weighted_segment).replace("\n", " ").split(' ')
+            to_tokenize = [x for x in to_tokenize if x != ""]
+            for word in to_tokenize:
+                #if we find an embedding, deal with the embedding
+                if word.startswith(self.embedding_identifier) and self.embedding_directory is not None:
+                    embedding_name = word[len(self.embedding_identifier):].strip('\n')
+                    embed, leftover = self._try_get_embedding(embedding_name)
+                    if embed is None:
+                        print(f"warning, embedding:{embedding_name} does not exist, ignoring")
+                    else:
+                        if len(embed.shape) == 1:
+                            tokens.append([(embed, weight)])
+                        else:
+                            tokens.append([(embed[x], weight) for x in range(embed.shape[0])])
+                    #if we accidentally have leftover text, continue parsing using leftover, else move on to next word
+                    if leftover != "":
+                        word = leftover
+                    else:
+                        continue
+                #parse word
+                tokens.append([(t, weight) for t in self.tokenizer(word)["input_ids"][self.tokens_start:-1]])
+
+        #reshape token array to CLIP input size
+        batched_tokens = []
+        batch = []
+        if self.start_token is not None:
+            batch.append((self.start_token, 1.0, 0))
+        batched_tokens.append(batch)
+        for i, t_group in enumerate(tokens):
+            #determine if we're going to try and keep the tokens in a single batch
+            is_large = len(t_group) >= self.max_word_length
+
+            while len(t_group) > 0:
+                if len(t_group) + len(batch) > self.max_length - 1:
+                    remaining_length = self.max_length - len(batch) - 1
+                    #break word in two and add end token
+                    if is_large:
+                        batch.extend([(t,w,i+1) for t,w in t_group[:remaining_length]])
+                        batch.append((self.end_token, 1.0, 0))
+                        t_group = t_group[remaining_length:]
+                    #add end token and pad
+                    else:
+                        batch.append((self.end_token, 1.0, 0))
+                        if self.pad_to_max_length:
+                            batch.extend([(pad_token, 1.0, 0)] * (remaining_length))
+                    #start new batch
+                    batch = []
+                    if self.start_token is not None:
+                        batch.append((self.start_token, 1.0, 0))
+                    batched_tokens.append(batch)
+                else:
+                    batch.extend([(t,w,i+1) for t,w in t_group])
+                    t_group = []
+
+        #fill last batch
+        batch.append((self.end_token, 1.0, 0))
+        if self.pad_to_max_length:
+            batch.extend([(pad_token, 1.0, 0)] * (self.max_length - len(batch)))
+
+        if not return_word_ids:
+            batched_tokens = [[(t, w) for t, w,_ in x] for x in batched_tokens]
+
+        return batched_tokens
+
+
+    def untokenize(self, token_weight_pair):
+        return list(map(lambda a: (a, self.inv_vocab[a[0]]), token_weight_pair))
+
+
+class SD1Tokenizer:
+    def __init__(self, embedding_directory=None, clip_name="l", tokenizer=SDTokenizer):
+        self.clip_name = clip_name
+        self.clip = "clip_{}".format(self.clip_name)
+        setattr(self, self.clip, tokenizer(embedding_directory=embedding_directory))
+
+    def tokenize_with_weights(self, text:str, return_word_ids=False):
+        out = {}
+        out[self.clip_name] = getattr(self, self.clip).tokenize_with_weights(text, return_word_ids)
+        return out
+
+    def untokenize(self, token_weight_pair):
+        return getattr(self, self.clip).untokenize(token_weight_pair)
+
+
+class SD1ClipModel(torch.nn.Module):
+    def __init__(self, device="cpu", dtype=None, clip_name="l", clip_model=SDClipModel, **kwargs):
+        super().__init__()
+        self.clip_name = clip_name
+        self.clip = "clip_{}".format(self.clip_name)
+        setattr(self, self.clip, clip_model(device=device, dtype=dtype, **kwargs))
+
+    def clip_layer(self, layer_idx):
+        getattr(self, self.clip).clip_layer(layer_idx)
+
+    def reset_clip_layer(self):
+        getattr(self, self.clip).reset_clip_layer()
+
+    def encode_token_weights(self, token_weight_pairs):
+        token_weight_pairs = token_weight_pairs[self.clip_name]
+        out, pooled = getattr(self, self.clip).encode_token_weights(token_weight_pairs)
+        return out, pooled
+
+    def load_sd(self, sd):
+        return getattr(self, self.clip).load_sd(sd)

backend/headless/fcbh/sd1_clip_config.json

@@ -0,0 +1,25 @@
+{
+  "_name_or_path": "openai/clip-vit-large-patch14",
+  "architectures": [
+    "CLIPTextModel"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 0,
+  "dropout": 0.0,
+  "eos_token_id": 2,
+  "hidden_act": "quick_gelu",
+  "hidden_size": 768,
+  "initializer_factor": 1.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 77,
+  "model_type": "clip_text_model",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 1,
+  "projection_dim": 768,
+  "torch_dtype": "float32",
+  "transformers_version": "4.24.0",
+  "vocab_size": 49408
+}

backend/headless/fcbh/sd1_tokenizer/special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<|startoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<|endoftext|>",
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

backend/headless/fcbh/sd1_tokenizer/tokenizer_config.json

@@ -0,0 +1,34 @@
+{
+  "add_prefix_space": false,
+  "bos_token": {
+    "__type": "AddedToken",
+    "content": "<|startoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "do_lower_case": true,
+  "eos_token": {
+    "__type": "AddedToken",
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "errors": "replace",
+  "model_max_length": 77,
+  "name_or_path": "openai/clip-vit-large-patch14",
+  "pad_token": "<|endoftext|>",
+  "special_tokens_map_file": "./special_tokens_map.json",
+  "tokenizer_class": "CLIPTokenizer",
+  "unk_token": {
+    "__type": "AddedToken",
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

backend/headless/fcbh/sd2_clip.py

@@ -0,0 +1,24 @@
+from fcbh import sd1_clip
+import torch
+import os
+
+class SD2ClipHModel(sd1_clip.SDClipModel):
+    def __init__(self, arch="ViT-H-14", device="cpu", max_length=77, freeze=True, layer="penultimate", layer_idx=None, textmodel_path=None, dtype=None):
+        if layer == "penultimate":
+            layer="hidden"
+            layer_idx=23
+
+        textmodel_json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "sd2_clip_config.json")
+        super().__init__(device=device, freeze=freeze, layer=layer, layer_idx=layer_idx, textmodel_json_config=textmodel_json_config, textmodel_path=textmodel_path, dtype=dtype, special_tokens={"start": 49406, "end": 49407, "pad": 0})
+
+class SD2ClipHTokenizer(sd1_clip.SDTokenizer):
+    def __init__(self, tokenizer_path=None, embedding_directory=None):
+        super().__init__(tokenizer_path, pad_with_end=False, embedding_directory=embedding_directory, embedding_size=1024)
+
+class SD2Tokenizer(sd1_clip.SD1Tokenizer):
+    def __init__(self, embedding_directory=None):
+        super().__init__(embedding_directory=embedding_directory, clip_name="h", tokenizer=SD2ClipHTokenizer)
+
+class SD2ClipModel(sd1_clip.SD1ClipModel):
+    def __init__(self, device="cpu", dtype=None, **kwargs):
+        super().__init__(device=device, dtype=dtype, clip_name="h", clip_model=SD2ClipHModel, **kwargs)

backend/headless/fcbh/sd2_clip_config.json

@@ -0,0 +1,23 @@
+{
+  "architectures": [
+    "CLIPTextModel"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 0,
+  "dropout": 0.0,
+  "eos_token_id": 2,
+  "hidden_act": "gelu",
+  "hidden_size": 1024,
+  "initializer_factor": 1.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 77,
+  "model_type": "clip_text_model",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
+  "pad_token_id": 1,
+  "projection_dim": 1024,
+  "torch_dtype": "float32",
+  "vocab_size": 49408
+}

backend/headless/fcbh/sdxl_clip.py

@@ -0,0 +1,66 @@
+from fcbh import sd1_clip
+import torch
+import os
+
+class SDXLClipG(sd1_clip.SDClipModel):
+    def __init__(self, device="cpu", max_length=77, freeze=True, layer="penultimate", layer_idx=None, textmodel_path=None, dtype=None):
+        if layer == "penultimate":
+            layer="hidden"
+            layer_idx=-2
+
+        textmodel_json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "clip_config_bigg.json")
+        super().__init__(device=device, freeze=freeze, layer=layer, layer_idx=layer_idx, textmodel_json_config=textmodel_json_config, textmodel_path=textmodel_path, dtype=dtype,
+                         special_tokens={"start": 49406, "end": 49407, "pad": 0}, layer_norm_hidden_state=False)
+
+    def load_sd(self, sd):
+        return super().load_sd(sd)
+
+class SDXLClipGTokenizer(sd1_clip.SDTokenizer):
+    def __init__(self, tokenizer_path=None, embedding_directory=None):
+        super().__init__(tokenizer_path, pad_with_end=False, embedding_directory=embedding_directory, embedding_size=1280, embedding_key='clip_g')
+
+
+class SDXLTokenizer:
+    def __init__(self, embedding_directory=None):
+        self.clip_l = sd1_clip.SDTokenizer(embedding_directory=embedding_directory)
+        self.clip_g = SDXLClipGTokenizer(embedding_directory=embedding_directory)
+
+    def tokenize_with_weights(self, text:str, return_word_ids=False):
+        out = {}
+        out["g"] = self.clip_g.tokenize_with_weights(text, return_word_ids)
+        out["l"] = self.clip_l.tokenize_with_weights(text, return_word_ids)
+        return out
+
+    def untokenize(self, token_weight_pair):
+        return self.clip_g.untokenize(token_weight_pair)
+
+class SDXLClipModel(torch.nn.Module):
+    def __init__(self, device="cpu", dtype=None):
+        super().__init__()
+        self.clip_l = sd1_clip.SDClipModel(layer="hidden", layer_idx=11, device=device, dtype=dtype, layer_norm_hidden_state=False)
+        self.clip_g = SDXLClipG(device=device, dtype=dtype)
+
+    def clip_layer(self, layer_idx):
+        self.clip_l.clip_layer(layer_idx)
+        self.clip_g.clip_layer(layer_idx)
+
+    def reset_clip_layer(self):
+        self.clip_g.reset_clip_layer()
+        self.clip_l.reset_clip_layer()
+
+    def encode_token_weights(self, token_weight_pairs):
+        token_weight_pairs_g = token_weight_pairs["g"]
+        token_weight_pairs_l = token_weight_pairs["l"]
+        g_out, g_pooled = self.clip_g.encode_token_weights(token_weight_pairs_g)
+        l_out, l_pooled = self.clip_l.encode_token_weights(token_weight_pairs_l)
+        return torch.cat([l_out, g_out], dim=-1), g_pooled
+
+    def load_sd(self, sd):
+        if "text_model.encoder.layers.30.mlp.fc1.weight" in sd:
+            return self.clip_g.load_sd(sd)
+        else:
+            return self.clip_l.load_sd(sd)
+
+class SDXLRefinerClipModel(sd1_clip.SD1ClipModel):
+    def __init__(self, device="cpu", dtype=None):
+        super().__init__(device=device, dtype=dtype, clip_name="g", clip_model=SDXLClipG)

backend/headless/fcbh/supported_models.py

@@ -0,0 +1,210 @@
+import torch
+from . import model_base
+from . import utils
+
+from . import sd1_clip
+from . import sd2_clip
+from . import sdxl_clip
+
+from . import supported_models_base
+from . import latent_formats
+
+from . import diffusers_convert
+
+class SD15(supported_models_base.BASE):
+    unet_config = {
+        "context_dim": 768,
+        "model_channels": 320,
+        "use_linear_in_transformer": False,
+        "adm_in_channels": None,
+    }
+
+    unet_extra_config = {
+        "num_heads": 8,
+        "num_head_channels": -1,
+    }
+
+    latent_format = latent_formats.SD15
+
+    def process_clip_state_dict(self, state_dict):
+        k = list(state_dict.keys())
+        for x in k:
+            if x.startswith("cond_stage_model.transformer.") and not x.startswith("cond_stage_model.transformer.text_model."):
+                y = x.replace("cond_stage_model.transformer.", "cond_stage_model.transformer.text_model.")
+                state_dict[y] = state_dict.pop(x)
+
+        if 'cond_stage_model.transformer.text_model.embeddings.position_ids' in state_dict:
+            ids = state_dict['cond_stage_model.transformer.text_model.embeddings.position_ids']
+            if ids.dtype == torch.float32:
+                state_dict['cond_stage_model.transformer.text_model.embeddings.position_ids'] = ids.round()
+
+        replace_prefix = {}
+        replace_prefix["cond_stage_model."] = "cond_stage_model.clip_l."
+        state_dict = utils.state_dict_prefix_replace(state_dict, replace_prefix)
+        return state_dict
+
+    def process_clip_state_dict_for_saving(self, state_dict):
+        replace_prefix = {"clip_l.": "cond_stage_model."}
+        return utils.state_dict_prefix_replace(state_dict, replace_prefix)
+
+    def clip_target(self):
+        return supported_models_base.ClipTarget(sd1_clip.SD1Tokenizer, sd1_clip.SD1ClipModel)
+
+class SD20(supported_models_base.BASE):
+    unet_config = {
+        "context_dim": 1024,
+        "model_channels": 320,
+        "use_linear_in_transformer": True,
+        "adm_in_channels": None,
+    }
+
+    latent_format = latent_formats.SD15
+
+    def model_type(self, state_dict, prefix=""):
+        if self.unet_config["in_channels"] == 4: #SD2.0 inpainting models are not v prediction
+            k = "{}output_blocks.11.1.transformer_blocks.0.norm1.bias".format(prefix)
+            out = state_dict[k]
+            if torch.std(out, unbiased=False) > 0.09: # not sure how well this will actually work. I guess we will find out.
+                return model_base.ModelType.V_PREDICTION
+        return model_base.ModelType.EPS
+
+    def process_clip_state_dict(self, state_dict):
+        state_dict = utils.transformers_convert(state_dict, "cond_stage_model.model.", "cond_stage_model.clip_h.transformer.text_model.", 24)
+        return state_dict
+
+    def process_clip_state_dict_for_saving(self, state_dict):
+        replace_prefix = {}
+        replace_prefix["clip_h"] = "cond_stage_model.model"
+        state_dict = utils.state_dict_prefix_replace(state_dict, replace_prefix)
+        state_dict = diffusers_convert.convert_text_enc_state_dict_v20(state_dict)
+        return state_dict
+
+    def clip_target(self):
+        return supported_models_base.ClipTarget(sd2_clip.SD2Tokenizer, sd2_clip.SD2ClipModel)
+
+class SD21UnclipL(SD20):
+    unet_config = {
+        "context_dim": 1024,
+        "model_channels": 320,
+        "use_linear_in_transformer": True,
+        "adm_in_channels": 1536,
+    }
+
+    clip_vision_prefix = "embedder.model.visual."
+    noise_aug_config = {"noise_schedule_config": {"timesteps": 1000, "beta_schedule": "squaredcos_cap_v2"}, "timestep_dim": 768}
+
+
+class SD21UnclipH(SD20):
+    unet_config = {
+        "context_dim": 1024,
+        "model_channels": 320,
+        "use_linear_in_transformer": True,
+        "adm_in_channels": 2048,
+    }
+
+    clip_vision_prefix = "embedder.model.visual."
+    noise_aug_config = {"noise_schedule_config": {"timesteps": 1000, "beta_schedule": "squaredcos_cap_v2"}, "timestep_dim": 1024}
+
+class SDXLRefiner(supported_models_base.BASE):
+    unet_config = {
+        "model_channels": 384,
+        "use_linear_in_transformer": True,
+        "context_dim": 1280,
+        "adm_in_channels": 2560,
+        "transformer_depth": [0, 0, 4, 4, 4, 4, 0, 0],
+    }
+
+    latent_format = latent_formats.SDXL
+
+    def get_model(self, state_dict, prefix="", device=None):
+        return model_base.SDXLRefiner(self, device=device)
+
+    def process_clip_state_dict(self, state_dict):
+        keys_to_replace = {}
+        replace_prefix = {}
+
+        state_dict = utils.transformers_convert(state_dict, "conditioner.embedders.0.model.", "cond_stage_model.clip_g.transformer.text_model.", 32)
+        keys_to_replace["conditioner.embedders.0.model.text_projection"] = "cond_stage_model.clip_g.text_projection"
+        keys_to_replace["conditioner.embedders.0.model.logit_scale"] = "cond_stage_model.clip_g.logit_scale"
+
+        state_dict = utils.state_dict_key_replace(state_dict, keys_to_replace)
+        return state_dict
+
+    def process_clip_state_dict_for_saving(self, state_dict):
+        replace_prefix = {}
+        state_dict_g = diffusers_convert.convert_text_enc_state_dict_v20(state_dict, "clip_g")
+        if "clip_g.transformer.text_model.embeddings.position_ids" in state_dict_g:
+            state_dict_g.pop("clip_g.transformer.text_model.embeddings.position_ids")
+        replace_prefix["clip_g"] = "conditioner.embedders.0.model"
+        state_dict_g = utils.state_dict_prefix_replace(state_dict_g, replace_prefix)
+        return state_dict_g
+
+    def clip_target(self):
+        return supported_models_base.ClipTarget(sdxl_clip.SDXLTokenizer, sdxl_clip.SDXLRefinerClipModel)
+
+class SDXL(supported_models_base.BASE):
+    unet_config = {
+        "model_channels": 320,
+        "use_linear_in_transformer": True,
+        "transformer_depth": [0, 0, 2, 2, 10, 10],
+        "context_dim": 2048,
+        "adm_in_channels": 2816
+    }
+
+    latent_format = latent_formats.SDXL
+
+    def model_type(self, state_dict, prefix=""):
+        if "v_pred" in state_dict:
+            return model_base.ModelType.V_PREDICTION
+        else:
+            return model_base.ModelType.EPS
+
+    def get_model(self, state_dict, prefix="", device=None):
+        out = model_base.SDXL(self, model_type=self.model_type(state_dict, prefix), device=device)
+        if self.inpaint_model():
+            out.set_inpaint()
+        return out
+
+    def process_clip_state_dict(self, state_dict):
+        keys_to_replace = {}
+        replace_prefix = {}
+
+        replace_prefix["conditioner.embedders.0.transformer.text_model"] = "cond_stage_model.clip_l.transformer.text_model"
+        state_dict = utils.transformers_convert(state_dict, "conditioner.embedders.1.model.", "cond_stage_model.clip_g.transformer.text_model.", 32)
+        keys_to_replace["conditioner.embedders.1.model.text_projection"] = "cond_stage_model.clip_g.text_projection"
+        keys_to_replace["conditioner.embedders.1.model.text_projection.weight"] = "cond_stage_model.clip_g.text_projection"
+        keys_to_replace["conditioner.embedders.1.model.logit_scale"] = "cond_stage_model.clip_g.logit_scale"
+
+        state_dict = utils.state_dict_prefix_replace(state_dict, replace_prefix)
+        state_dict = utils.state_dict_key_replace(state_dict, keys_to_replace)
+        return state_dict
+
+    def process_clip_state_dict_for_saving(self, state_dict):
+        replace_prefix = {}
+        keys_to_replace = {}
+        state_dict_g = diffusers_convert.convert_text_enc_state_dict_v20(state_dict, "clip_g")
+        if "clip_g.transformer.text_model.embeddings.position_ids" in state_dict_g:
+            state_dict_g.pop("clip_g.transformer.text_model.embeddings.position_ids")
+        for k in state_dict:
+            if k.startswith("clip_l"):
+                state_dict_g[k] = state_dict[k]
+
+        replace_prefix["clip_g"] = "conditioner.embedders.1.model"
+        replace_prefix["clip_l"] = "conditioner.embedders.0"
+        state_dict_g = utils.state_dict_prefix_replace(state_dict_g, replace_prefix)
+        return state_dict_g
+
+    def clip_target(self):
+        return supported_models_base.ClipTarget(sdxl_clip.SDXLTokenizer, sdxl_clip.SDXLClipModel)
+
+class SSD1B(SDXL):
+    unet_config = {
+        "model_channels": 320,
+        "use_linear_in_transformer": True,
+        "transformer_depth": [0, 0, 2, 2, 4, 4],
+        "context_dim": 2048,
+        "adm_in_channels": 2816
+    }
+
+
+models = [SD15, SD20, SD21UnclipL, SD21UnclipH, SDXLRefiner, SDXL, SSD1B]

backend/headless/fcbh/supported_models_base.py

@@ -0,0 +1,67 @@
+import torch
+from . import model_base
+from . import utils
+from . import latent_formats
+
+class ClipTarget:
+    def __init__(self, tokenizer, clip):
+        self.clip = clip
+        self.tokenizer = tokenizer
+        self.params = {}
+
+class BASE:
+    unet_config = {}
+    unet_extra_config = {
+        "num_heads": -1,
+        "num_head_channels": 64,
+    }
+
+    clip_prefix = []
+    clip_vision_prefix = None
+    noise_aug_config = None
+    beta_schedule = "linear"
+    latent_format = latent_formats.LatentFormat
+
+    @classmethod
+    def matches(s, unet_config):
+        for k in s.unet_config:
+            if s.unet_config[k] != unet_config[k]:
+                return False
+        return True
+
+    def model_type(self, state_dict, prefix=""):
+        return model_base.ModelType.EPS
+
+    def inpaint_model(self):
+        return self.unet_config["in_channels"] > 4
+
+    def __init__(self, unet_config):
+        self.unet_config = unet_config
+        self.latent_format = self.latent_format()
+        for x in self.unet_extra_config:
+            self.unet_config[x] = self.unet_extra_config[x]
+
+    def get_model(self, state_dict, prefix="", device=None):
+        if self.noise_aug_config is not None:
+            out = model_base.SD21UNCLIP(self, self.noise_aug_config, model_type=self.model_type(state_dict, prefix), device=device)
+        else:
+            out = model_base.BaseModel(self, model_type=self.model_type(state_dict, prefix), device=device)
+        if self.inpaint_model():
+            out.set_inpaint()
+        return out
+
+    def process_clip_state_dict(self, state_dict):
+        return state_dict
+
+    def process_clip_state_dict_for_saving(self, state_dict):
+        replace_prefix = {"": "cond_stage_model."}
+        return utils.state_dict_prefix_replace(state_dict, replace_prefix)
+
+    def process_unet_state_dict_for_saving(self, state_dict):
+        replace_prefix = {"": "model.diffusion_model."}
+        return utils.state_dict_prefix_replace(state_dict, replace_prefix)
+
+    def process_vae_state_dict_for_saving(self, state_dict):
+        replace_prefix = {"": "first_stage_model."}
+        return utils.state_dict_prefix_replace(state_dict, replace_prefix)
+

backend/headless/fcbh/t2i_adapter/adapter.py

@@ -0,0 +1,293 @@
+#taken from https://github.com/TencentARC/T2I-Adapter
+import torch
+import torch.nn as nn
+from collections import OrderedDict
+
+
+def conv_nd(dims, *args, **kwargs):
+    """
+    Create a 1D, 2D, or 3D convolution module.
+    """
+    if dims == 1:
+        return nn.Conv1d(*args, **kwargs)
+    elif dims == 2:
+        return nn.Conv2d(*args, **kwargs)
+    elif dims == 3:
+        return nn.Conv3d(*args, **kwargs)
+    raise ValueError(f"unsupported dimensions: {dims}")
+
+
+def avg_pool_nd(dims, *args, **kwargs):
+    """
+    Create a 1D, 2D, or 3D average pooling module.
+    """
+    if dims == 1:
+        return nn.AvgPool1d(*args, **kwargs)
+    elif dims == 2:
+        return nn.AvgPool2d(*args, **kwargs)
+    elif dims == 3:
+        return nn.AvgPool3d(*args, **kwargs)
+    raise ValueError(f"unsupported dimensions: {dims}")
+
+
+class Downsample(nn.Module):
+    """
+    A downsampling layer with an optional convolution.
+    :param channels: channels in the inputs and outputs.
+    :param use_conv: a bool determining if a convolution is applied.
+    :param dims: determines if the signal is 1D, 2D, or 3D. If 3D, then
+                 downsampling occurs in the inner-two dimensions.
+    """
+
+    def __init__(self, channels, use_conv, dims=2, out_channels=None, padding=1):
+        super().__init__()
+        self.channels = channels
+        self.out_channels = out_channels or channels
+        self.use_conv = use_conv
+        self.dims = dims
+        stride = 2 if dims != 3 else (1, 2, 2)
+        if use_conv:
+            self.op = conv_nd(
+                dims, self.channels, self.out_channels, 3, stride=stride, padding=padding
+            )
+        else:
+            assert self.channels == self.out_channels
+            self.op = avg_pool_nd(dims, kernel_size=stride, stride=stride)
+
+    def forward(self, x):
+        assert x.shape[1] == self.channels
+        if not self.use_conv:
+            padding = [x.shape[2] % 2, x.shape[3] % 2]
+            self.op.padding = padding
+
+        x = self.op(x)
+        return x
+
+
+class ResnetBlock(nn.Module):
+    def __init__(self, in_c, out_c, down, ksize=3, sk=False, use_conv=True):
+        super().__init__()
+        ps = ksize // 2
+        if in_c != out_c or sk == False:
+            self.in_conv = nn.Conv2d(in_c, out_c, ksize, 1, ps)
+        else:
+            # print('n_in')
+            self.in_conv = None
+        self.block1 = nn.Conv2d(out_c, out_c, 3, 1, 1)
+        self.act = nn.ReLU()
+        self.block2 = nn.Conv2d(out_c, out_c, ksize, 1, ps)
+        if sk == False:
+            self.skep = nn.Conv2d(in_c, out_c, ksize, 1, ps)
+        else:
+            self.skep = None
+
+        self.down = down
+        if self.down == True:
+            self.down_opt = Downsample(in_c, use_conv=use_conv)
+
+    def forward(self, x):
+        if self.down == True:
+            x = self.down_opt(x)
+        if self.in_conv is not None:  # edit
+            x = self.in_conv(x)
+
+        h = self.block1(x)
+        h = self.act(h)
+        h = self.block2(h)
+        if self.skep is not None:
+            return h + self.skep(x)
+        else:
+            return h + x
+
+
+class Adapter(nn.Module):
+    def __init__(self, channels=[320, 640, 1280, 1280], nums_rb=3, cin=64, ksize=3, sk=False, use_conv=True, xl=True):
+        super(Adapter, self).__init__()
+        self.unshuffle_amount = 8
+        resblock_no_downsample = []
+        resblock_downsample = [3, 2, 1]
+        self.xl = xl
+        if self.xl:
+            self.unshuffle_amount = 16
+            resblock_no_downsample = [1]
+            resblock_downsample = [2]
+
+        self.input_channels = cin // (self.unshuffle_amount * self.unshuffle_amount)
+        self.unshuffle = nn.PixelUnshuffle(self.unshuffle_amount)
+        self.channels = channels
+        self.nums_rb = nums_rb
+        self.body = []
+        for i in range(len(channels)):
+            for j in range(nums_rb):
+                if (i in resblock_downsample) and (j == 0):
+                    self.body.append(
+                        ResnetBlock(channels[i - 1], channels[i], down=True, ksize=ksize, sk=sk, use_conv=use_conv))
+                elif (i in resblock_no_downsample) and (j == 0):
+                    self.body.append(
+                        ResnetBlock(channels[i - 1], channels[i], down=False, ksize=ksize, sk=sk, use_conv=use_conv))
+                else:
+                    self.body.append(
+                        ResnetBlock(channels[i], channels[i], down=False, ksize=ksize, sk=sk, use_conv=use_conv))
+        self.body = nn.ModuleList(self.body)
+        self.conv_in = nn.Conv2d(cin, channels[0], 3, 1, 1)
+
+    def forward(self, x):
+        # unshuffle
+        x = self.unshuffle(x)
+        # extract features
+        features = []
+        x = self.conv_in(x)
+        for i in range(len(self.channels)):
+            for j in range(self.nums_rb):
+                idx = i * self.nums_rb + j
+                x = self.body[idx](x)
+            if self.xl:
+                features.append(None)
+                if i == 0:
+                    features.append(None)
+                    features.append(None)
+                if i == 2:
+                    features.append(None)
+            else:
+                features.append(None)
+                features.append(None)
+            features.append(x)
+
+        return features
+
+
+class LayerNorm(nn.LayerNorm):
+    """Subclass torch's LayerNorm to handle fp16."""
+
+    def forward(self, x: torch.Tensor):
+        orig_type = x.dtype
+        ret = super().forward(x.type(torch.float32))
+        return ret.type(orig_type)
+
+
+class QuickGELU(nn.Module):
+
+    def forward(self, x: torch.Tensor):
+        return x * torch.sigmoid(1.702 * x)
+
+
+class ResidualAttentionBlock(nn.Module):
+
+    def __init__(self, d_model: int, n_head: int, attn_mask: torch.Tensor = None):
+        super().__init__()
+
+        self.attn = nn.MultiheadAttention(d_model, n_head)
+        self.ln_1 = LayerNorm(d_model)
+        self.mlp = nn.Sequential(
+            OrderedDict([("c_fc", nn.Linear(d_model, d_model * 4)), ("gelu", QuickGELU()),
+                         ("c_proj", nn.Linear(d_model * 4, d_model))]))
+        self.ln_2 = LayerNorm(d_model)
+        self.attn_mask = attn_mask
+
+    def attention(self, x: torch.Tensor):
+        self.attn_mask = self.attn_mask.to(dtype=x.dtype, device=x.device) if self.attn_mask is not None else None
+        return self.attn(x, x, x, need_weights=False, attn_mask=self.attn_mask)[0]
+
+    def forward(self, x: torch.Tensor):
+        x = x + self.attention(self.ln_1(x))
+        x = x + self.mlp(self.ln_2(x))
+        return x
+
+
+class StyleAdapter(nn.Module):
+
+    def __init__(self, width=1024, context_dim=768, num_head=8, n_layes=3, num_token=4):
+        super().__init__()
+
+        scale = width ** -0.5
+        self.transformer_layes = nn.Sequential(*[ResidualAttentionBlock(width, num_head) for _ in range(n_layes)])
+        self.num_token = num_token
+        self.style_embedding = nn.Parameter(torch.randn(1, num_token, width) * scale)
+        self.ln_post = LayerNorm(width)
+        self.ln_pre = LayerNorm(width)
+        self.proj = nn.Parameter(scale * torch.randn(width, context_dim))
+
+    def forward(self, x):
+        # x shape [N, HW+1, C]
+        style_embedding = self.style_embedding + torch.zeros(
+            (x.shape[0], self.num_token, self.style_embedding.shape[-1]), device=x.device)
+        x = torch.cat([x, style_embedding], dim=1)
+        x = self.ln_pre(x)
+        x = x.permute(1, 0, 2)  # NLD -> LND
+        x = self.transformer_layes(x)
+        x = x.permute(1, 0, 2)  # LND -> NLD
+
+        x = self.ln_post(x[:, -self.num_token:, :])
+        x = x @ self.proj
+
+        return x
+
+
+class ResnetBlock_light(nn.Module):
+    def __init__(self, in_c):
+        super().__init__()
+        self.block1 = nn.Conv2d(in_c, in_c, 3, 1, 1)
+        self.act = nn.ReLU()
+        self.block2 = nn.Conv2d(in_c, in_c, 3, 1, 1)
+
+    def forward(self, x):
+        h = self.block1(x)
+        h = self.act(h)
+        h = self.block2(h)
+
+        return h + x
+
+
+class extractor(nn.Module):
+    def __init__(self, in_c, inter_c, out_c, nums_rb, down=False):
+        super().__init__()
+        self.in_conv = nn.Conv2d(in_c, inter_c, 1, 1, 0)
+        self.body = []
+        for _ in range(nums_rb):
+            self.body.append(ResnetBlock_light(inter_c))
+        self.body = nn.Sequential(*self.body)
+        self.out_conv = nn.Conv2d(inter_c, out_c, 1, 1, 0)
+        self.down = down
+        if self.down == True:
+            self.down_opt = Downsample(in_c, use_conv=False)
+
+    def forward(self, x):
+        if self.down == True:
+            x = self.down_opt(x)
+        x = self.in_conv(x)
+        x = self.body(x)
+        x = self.out_conv(x)
+
+        return x
+
+
+class Adapter_light(nn.Module):
+    def __init__(self, channels=[320, 640, 1280, 1280], nums_rb=3, cin=64):
+        super(Adapter_light, self).__init__()
+        self.unshuffle_amount = 8
+        self.unshuffle = nn.PixelUnshuffle(self.unshuffle_amount)
+        self.input_channels = cin // (self.unshuffle_amount * self.unshuffle_amount)
+        self.channels = channels
+        self.nums_rb = nums_rb
+        self.body = []
+        self.xl = False
+
+        for i in range(len(channels)):
+            if i == 0:
+                self.body.append(extractor(in_c=cin, inter_c=channels[i]//4, out_c=channels[i], nums_rb=nums_rb, down=False))
+            else:
+                self.body.append(extractor(in_c=channels[i-1], inter_c=channels[i]//4, out_c=channels[i], nums_rb=nums_rb, down=True))
+        self.body = nn.ModuleList(self.body)
+
+    def forward(self, x):
+        # unshuffle
+        x = self.unshuffle(x)
+        # extract features
+        features = []
+        for i in range(len(self.channels)):
+            x = self.body[i](x)
+            features.append(None)
+            features.append(None)
+            features.append(x)
+
+        return features

backend/headless/fcbh/taesd/taesd.py

@@ -0,0 +1,67 @@
+#!/usr/bin/env python3
+"""
+Tiny AutoEncoder for Stable Diffusion
+(DNN for encoding / decoding SD's latent space)
+"""
+import torch
+import torch.nn as nn
+
+import fcbh.utils
+
+def conv(n_in, n_out, **kwargs):
+    return nn.Conv2d(n_in, n_out, 3, padding=1, **kwargs)
+
+class Clamp(nn.Module):
+    def forward(self, x):
+        return torch.tanh(x / 3) * 3
+
+class Block(nn.Module):
+    def __init__(self, n_in, n_out):
+        super().__init__()
+        self.conv = nn.Sequential(conv(n_in, n_out), nn.ReLU(), conv(n_out, n_out), nn.ReLU(), conv(n_out, n_out))
+        self.skip = nn.Conv2d(n_in, n_out, 1, bias=False) if n_in != n_out else nn.Identity()
+        self.fuse = nn.ReLU()
+    def forward(self, x):
+        return self.fuse(self.conv(x) + self.skip(x))
+
+def Encoder():
+    return nn.Sequential(
+        conv(3, 64), Block(64, 64),
+        conv(64, 64, stride=2, bias=False), Block(64, 64), Block(64, 64), Block(64, 64),
+        conv(64, 64, stride=2, bias=False), Block(64, 64), Block(64, 64), Block(64, 64),
+        conv(64, 64, stride=2, bias=False), Block(64, 64), Block(64, 64), Block(64, 64),
+        conv(64, 4),
+    )
+
+def Decoder():
+    return nn.Sequential(
+        Clamp(), conv(4, 64), nn.ReLU(),
+        Block(64, 64), Block(64, 64), Block(64, 64), nn.Upsample(scale_factor=2), conv(64, 64, bias=False),
+        Block(64, 64), Block(64, 64), Block(64, 64), nn.Upsample(scale_factor=2), conv(64, 64, bias=False),
+        Block(64, 64), Block(64, 64), Block(64, 64), nn.Upsample(scale_factor=2), conv(64, 64, bias=False),
+        Block(64, 64), conv(64, 3),
+    )
+
+class TAESD(nn.Module):
+    latent_magnitude = 3
+    latent_shift = 0.5
+
+    def __init__(self, encoder_path="taesd_encoder.pth", decoder_path="taesd_decoder.pth"):
+        """Initialize pretrained TAESD on the given device from the given checkpoints."""
+        super().__init__()
+        self.encoder = Encoder()
+        self.decoder = Decoder()
+        if encoder_path is not None:
+            self.encoder.load_state_dict(fcbh.utils.load_torch_file(encoder_path, safe_load=True))
+        if decoder_path is not None:
+            self.decoder.load_state_dict(fcbh.utils.load_torch_file(decoder_path, safe_load=True))
+
+    @staticmethod
+    def scale_latents(x):
+        """raw latents -> [0, 1]"""
+        return x.div(2 * TAESD.latent_magnitude).add(TAESD.latent_shift).clamp(0, 1)
+
+    @staticmethod
+    def unscale_latents(x):
+        """[0, 1] -> raw latents"""
+        return x.sub(TAESD.latent_shift).mul(2 * TAESD.latent_magnitude)

backend/headless/fcbh/utils.py

@@ -0,0 +1,431 @@
+import torch
+import math
+import struct
+import fcbh.checkpoint_pickle
+import safetensors.torch
+import numpy as np
+from PIL import Image
+
+def load_torch_file(ckpt, safe_load=False, device=None):
+    if device is None:
+        device = torch.device("cpu")
+    if ckpt.lower().endswith(".safetensors"):
+        sd = safetensors.torch.load_file(ckpt, device=device.type)
+    else:
+        if safe_load:
+            if not 'weights_only' in torch.load.__code__.co_varnames:
+                print("Warning torch.load doesn't support weights_only on this pytorch version, loading unsafely.")
+                safe_load = False
+        if safe_load:
+            pl_sd = torch.load(ckpt, map_location=device, weights_only=True)
+        else:
+            pl_sd = torch.load(ckpt, map_location=device, pickle_module=fcbh.checkpoint_pickle)
+        if "global_step" in pl_sd:
+            print(f"Global Step: {pl_sd['global_step']}")
+        if "state_dict" in pl_sd:
+            sd = pl_sd["state_dict"]
+        else:
+            sd = pl_sd
+    return sd
+
+def save_torch_file(sd, ckpt, metadata=None):
+    if metadata is not None:
+        safetensors.torch.save_file(sd, ckpt, metadata=metadata)
+    else:
+        safetensors.torch.save_file(sd, ckpt)
+
+def calculate_parameters(sd, prefix=""):
+    params = 0
+    for k in sd.keys():
+        if k.startswith(prefix):
+            params += sd[k].nelement()
+    return params
+
+def state_dict_key_replace(state_dict, keys_to_replace):
+    for x in keys_to_replace:
+        if x in state_dict:
+            state_dict[keys_to_replace[x]] = state_dict.pop(x)
+    return state_dict
+
+def state_dict_prefix_replace(state_dict, replace_prefix, filter_keys=False):
+    if filter_keys:
+        out = {}
+    else:
+        out = state_dict
+    for rp in replace_prefix:
+        replace = list(map(lambda a: (a, "{}{}".format(replace_prefix[rp], a[len(rp):])), filter(lambda a: a.startswith(rp), state_dict.keys())))
+        for x in replace:
+            w = state_dict.pop(x[0])
+            out[x[1]] = w
+    return out
+
+
+def transformers_convert(sd, prefix_from, prefix_to, number):
+    keys_to_replace = {
+        "{}positional_embedding": "{}embeddings.position_embedding.weight",
+        "{}token_embedding.weight": "{}embeddings.token_embedding.weight",
+        "{}ln_final.weight": "{}final_layer_norm.weight",
+        "{}ln_final.bias": "{}final_layer_norm.bias",
+    }
+
+    for k in keys_to_replace:
+        x = k.format(prefix_from)
+        if x in sd:
+            sd[keys_to_replace[k].format(prefix_to)] = sd.pop(x)
+
+    resblock_to_replace = {
+        "ln_1": "layer_norm1",
+        "ln_2": "layer_norm2",
+        "mlp.c_fc": "mlp.fc1",
+        "mlp.c_proj": "mlp.fc2",
+        "attn.out_proj": "self_attn.out_proj",
+    }
+
+    for resblock in range(number):
+        for x in resblock_to_replace:
+            for y in ["weight", "bias"]:
+                k = "{}transformer.resblocks.{}.{}.{}".format(prefix_from, resblock, x, y)
+                k_to = "{}encoder.layers.{}.{}.{}".format(prefix_to, resblock, resblock_to_replace[x], y)
+                if k in sd:
+                    sd[k_to] = sd.pop(k)
+
+        for y in ["weight", "bias"]:
+            k_from = "{}transformer.resblocks.{}.attn.in_proj_{}".format(prefix_from, resblock, y)
+            if k_from in sd:
+                weights = sd.pop(k_from)
+                shape_from = weights.shape[0] // 3
+                for x in range(3):
+                    p = ["self_attn.q_proj", "self_attn.k_proj", "self_attn.v_proj"]
+                    k_to = "{}encoder.layers.{}.{}.{}".format(prefix_to, resblock, p[x], y)
+                    sd[k_to] = weights[shape_from*x:shape_from*(x + 1)]
+    return sd
+
+UNET_MAP_ATTENTIONS = {
+    "proj_in.weight",
+    "proj_in.bias",
+    "proj_out.weight",
+    "proj_out.bias",
+    "norm.weight",
+    "norm.bias",
+}
+
+TRANSFORMER_BLOCKS = {
+    "norm1.weight",
+    "norm1.bias",
+    "norm2.weight",
+    "norm2.bias",
+    "norm3.weight",
+    "norm3.bias",
+    "attn1.to_q.weight",
+    "attn1.to_k.weight",
+    "attn1.to_v.weight",
+    "attn1.to_out.0.weight",
+    "attn1.to_out.0.bias",
+    "attn2.to_q.weight",
+    "attn2.to_k.weight",
+    "attn2.to_v.weight",
+    "attn2.to_out.0.weight",
+    "attn2.to_out.0.bias",
+    "ff.net.0.proj.weight",
+    "ff.net.0.proj.bias",
+    "ff.net.2.weight",
+    "ff.net.2.bias",
+}
+
+UNET_MAP_RESNET = {
+    "in_layers.2.weight": "conv1.weight",
+    "in_layers.2.bias": "conv1.bias",
+    "emb_layers.1.weight": "time_emb_proj.weight",
+    "emb_layers.1.bias": "time_emb_proj.bias",
+    "out_layers.3.weight": "conv2.weight",
+    "out_layers.3.bias": "conv2.bias",
+    "skip_connection.weight": "conv_shortcut.weight",
+    "skip_connection.bias": "conv_shortcut.bias",
+    "in_layers.0.weight": "norm1.weight",
+    "in_layers.0.bias": "norm1.bias",
+    "out_layers.0.weight": "norm2.weight",
+    "out_layers.0.bias": "norm2.bias",
+}
+
+UNET_MAP_BASIC = {
+    ("label_emb.0.0.weight", "class_embedding.linear_1.weight"),
+    ("label_emb.0.0.bias", "class_embedding.linear_1.bias"),
+    ("label_emb.0.2.weight", "class_embedding.linear_2.weight"),
+    ("label_emb.0.2.bias", "class_embedding.linear_2.bias"),
+    ("label_emb.0.0.weight", "add_embedding.linear_1.weight"),
+    ("label_emb.0.0.bias", "add_embedding.linear_1.bias"),
+    ("label_emb.0.2.weight", "add_embedding.linear_2.weight"),
+    ("label_emb.0.2.bias", "add_embedding.linear_2.bias"),
+    ("input_blocks.0.0.weight", "conv_in.weight"),
+    ("input_blocks.0.0.bias", "conv_in.bias"),
+    ("out.0.weight", "conv_norm_out.weight"),
+    ("out.0.bias", "conv_norm_out.bias"),
+    ("out.2.weight", "conv_out.weight"),
+    ("out.2.bias", "conv_out.bias"),
+    ("time_embed.0.weight", "time_embedding.linear_1.weight"),
+    ("time_embed.0.bias", "time_embedding.linear_1.bias"),
+    ("time_embed.2.weight", "time_embedding.linear_2.weight"),
+    ("time_embed.2.bias", "time_embedding.linear_2.bias")
+}
+
+def unet_to_diffusers(unet_config):
+    num_res_blocks = unet_config["num_res_blocks"]
+    channel_mult = unet_config["channel_mult"]
+    transformer_depth = unet_config["transformer_depth"][:]
+    transformer_depth_output = unet_config["transformer_depth_output"][:]
+    num_blocks = len(channel_mult)
+
+    transformers_mid = unet_config.get("transformer_depth_middle", None)
+
+    diffusers_unet_map = {}
+    for x in range(num_blocks):
+        n = 1 + (num_res_blocks[x] + 1) * x
+        for i in range(num_res_blocks[x]):
+            for b in UNET_MAP_RESNET:
+                diffusers_unet_map["down_blocks.{}.resnets.{}.{}".format(x, i, UNET_MAP_RESNET[b])] = "input_blocks.{}.0.{}".format(n, b)
+            num_transformers = transformer_depth.pop(0)
+            if num_transformers > 0:
+                for b in UNET_MAP_ATTENTIONS:
+                    diffusers_unet_map["down_blocks.{}.attentions.{}.{}".format(x, i, b)] = "input_blocks.{}.1.{}".format(n, b)
+                for t in range(num_transformers):
+                    for b in TRANSFORMER_BLOCKS:
+                        diffusers_unet_map["down_blocks.{}.attentions.{}.transformer_blocks.{}.{}".format(x, i, t, b)] = "input_blocks.{}.1.transformer_blocks.{}.{}".format(n, t, b)
+            n += 1
+        for k in ["weight", "bias"]:
+            diffusers_unet_map["down_blocks.{}.downsamplers.0.conv.{}".format(x, k)] = "input_blocks.{}.0.op.{}".format(n, k)
+
+    i = 0
+    for b in UNET_MAP_ATTENTIONS:
+        diffusers_unet_map["mid_block.attentions.{}.{}".format(i, b)] = "middle_block.1.{}".format(b)
+    for t in range(transformers_mid):
+        for b in TRANSFORMER_BLOCKS:
+            diffusers_unet_map["mid_block.attentions.{}.transformer_blocks.{}.{}".format(i, t, b)] = "middle_block.1.transformer_blocks.{}.{}".format(t, b)
+
+    for i, n in enumerate([0, 2]):
+        for b in UNET_MAP_RESNET:
+            diffusers_unet_map["mid_block.resnets.{}.{}".format(i, UNET_MAP_RESNET[b])] = "middle_block.{}.{}".format(n, b)
+
+    num_res_blocks = list(reversed(num_res_blocks))
+    for x in range(num_blocks):
+        n = (num_res_blocks[x] + 1) * x
+        l = num_res_blocks[x] + 1
+        for i in range(l):
+            c = 0
+            for b in UNET_MAP_RESNET:
+                diffusers_unet_map["up_blocks.{}.resnets.{}.{}".format(x, i, UNET_MAP_RESNET[b])] = "output_blocks.{}.0.{}".format(n, b)
+            c += 1
+            num_transformers = transformer_depth_output.pop()
+            if num_transformers > 0:
+                c += 1
+                for b in UNET_MAP_ATTENTIONS:
+                    diffusers_unet_map["up_blocks.{}.attentions.{}.{}".format(x, i, b)] = "output_blocks.{}.1.{}".format(n, b)
+                for t in range(num_transformers):
+                    for b in TRANSFORMER_BLOCKS:
+                        diffusers_unet_map["up_blocks.{}.attentions.{}.transformer_blocks.{}.{}".format(x, i, t, b)] = "output_blocks.{}.1.transformer_blocks.{}.{}".format(n, t, b)
+            if i == l - 1:
+                for k in ["weight", "bias"]:
+                    diffusers_unet_map["up_blocks.{}.upsamplers.0.conv.{}".format(x, k)] = "output_blocks.{}.{}.conv.{}".format(n, c, k)
+            n += 1
+
+    for k in UNET_MAP_BASIC:
+        diffusers_unet_map[k[1]] = k[0]
+
+    return diffusers_unet_map
+
+def repeat_to_batch_size(tensor, batch_size):
+    if tensor.shape[0] > batch_size:
+        return tensor[:batch_size]
+    elif tensor.shape[0] < batch_size:
+        return tensor.repeat([math.ceil(batch_size / tensor.shape[0])] + [1] * (len(tensor.shape) - 1))[:batch_size]
+    return tensor
+
+def convert_sd_to(state_dict, dtype):
+    keys = list(state_dict.keys())
+    for k in keys:
+        state_dict[k] = state_dict[k].to(dtype)
+    return state_dict
+
+def safetensors_header(safetensors_path, max_size=100*1024*1024):
+    with open(safetensors_path, "rb") as f:
+        header = f.read(8)
+        length_of_header = struct.unpack('<Q', header)[0]
+        if length_of_header > max_size:
+            return None
+        return f.read(length_of_header)
+
+def set_attr(obj, attr, value):
+    attrs = attr.split(".")
+    for name in attrs[:-1]:
+        obj = getattr(obj, name)
+    prev = getattr(obj, attrs[-1])
+    setattr(obj, attrs[-1], torch.nn.Parameter(value))
+    del prev
+
+def get_attr(obj, attr):
+    attrs = attr.split(".")
+    for name in attrs:
+        obj = getattr(obj, name)
+    return obj
+
+def bislerp(samples, width, height):
+    def slerp(b1, b2, r):
+        '''slerps batches b1, b2 according to ratio r, batches should be flat e.g. NxC'''
+        
+        c = b1.shape[-1]
+
+        #norms
+        b1_norms = torch.norm(b1, dim=-1, keepdim=True)
+        b2_norms = torch.norm(b2, dim=-1, keepdim=True)
+
+        #normalize
+        b1_normalized = b1 / b1_norms
+        b2_normalized = b2 / b2_norms
+
+        #zero when norms are zero
+        b1_normalized[b1_norms.expand(-1,c) == 0.0] = 0.0
+        b2_normalized[b2_norms.expand(-1,c) == 0.0] = 0.0
+
+        #slerp
+        dot = (b1_normalized*b2_normalized).sum(1)
+        omega = torch.acos(dot)
+        so = torch.sin(omega)
+
+        #technically not mathematically correct, but more pleasing?
+        res = (torch.sin((1.0-r.squeeze(1))*omega)/so).unsqueeze(1)*b1_normalized + (torch.sin(r.squeeze(1)*omega)/so).unsqueeze(1) * b2_normalized
+        res *= (b1_norms * (1.0-r) + b2_norms * r).expand(-1,c)
+
+        #edge cases for same or polar opposites
+        res[dot > 1 - 1e-5] = b1[dot > 1 - 1e-5] 
+        res[dot < 1e-5 - 1] = (b1 * (1.0-r) + b2 * r)[dot < 1e-5 - 1]
+        return res
+    
+    def generate_bilinear_data(length_old, length_new):
+        coords_1 = torch.arange(length_old).reshape((1,1,1,-1)).to(torch.float32)
+        coords_1 = torch.nn.functional.interpolate(coords_1, size=(1, length_new), mode="bilinear")
+        ratios = coords_1 - coords_1.floor()
+        coords_1 = coords_1.to(torch.int64)
+        
+        coords_2 = torch.arange(length_old).reshape((1,1,1,-1)).to(torch.float32) + 1
+        coords_2[:,:,:,-1] -= 1
+        coords_2 = torch.nn.functional.interpolate(coords_2, size=(1, length_new), mode="bilinear")
+        coords_2 = coords_2.to(torch.int64)
+        return ratios, coords_1, coords_2
+    
+    n,c,h,w = samples.shape
+    h_new, w_new = (height, width)
+    
+    #linear w
+    ratios, coords_1, coords_2 = generate_bilinear_data(w, w_new)
+    coords_1 = coords_1.expand((n, c, h, -1))
+    coords_2 = coords_2.expand((n, c, h, -1))
+    ratios = ratios.expand((n, 1, h, -1))
+
+    pass_1 = samples.gather(-1,coords_1).movedim(1, -1).reshape((-1,c))
+    pass_2 = samples.gather(-1,coords_2).movedim(1, -1).reshape((-1,c))
+    ratios = ratios.movedim(1, -1).reshape((-1,1))
+
+    result = slerp(pass_1, pass_2, ratios)
+    result = result.reshape(n, h, w_new, c).movedim(-1, 1)
+
+    #linear h
+    ratios, coords_1, coords_2 = generate_bilinear_data(h, h_new)
+    coords_1 = coords_1.reshape((1,1,-1,1)).expand((n, c, -1, w_new))
+    coords_2 = coords_2.reshape((1,1,-1,1)).expand((n, c, -1, w_new))
+    ratios = ratios.reshape((1,1,-1,1)).expand((n, 1, -1, w_new))
+
+    pass_1 = result.gather(-2,coords_1).movedim(1, -1).reshape((-1,c))
+    pass_2 = result.gather(-2,coords_2).movedim(1, -1).reshape((-1,c))
+    ratios = ratios.movedim(1, -1).reshape((-1,1))
+
+    result = slerp(pass_1, pass_2, ratios)
+    result = result.reshape(n, h_new, w_new, c).movedim(-1, 1)
+    return result
+
+def lanczos(samples, width, height):
+    images = [Image.fromarray(np.clip(255. * image.movedim(0, -1).cpu().numpy(), 0, 255).astype(np.uint8)) for image in samples]
+    images = [image.resize((width, height), resample=Image.Resampling.LANCZOS) for image in images]
+    images = [torch.from_numpy(np.array(image).astype(np.float32) / 255.0).movedim(-1, 0) for image in images]
+    result = torch.stack(images)
+    return result
+
+def common_upscale(samples, width, height, upscale_method, crop):
+        if crop == "center":
+            old_width = samples.shape[3]
+            old_height = samples.shape[2]
+            old_aspect = old_width / old_height
+            new_aspect = width / height
+            x = 0
+            y = 0
+            if old_aspect > new_aspect:
+                x = round((old_width - old_width * (new_aspect / old_aspect)) / 2)
+            elif old_aspect < new_aspect:
+                y = round((old_height - old_height * (old_aspect / new_aspect)) / 2)
+            s = samples[:,:,y:old_height-y,x:old_width-x]
+        else:
+            s = samples
+
+        if upscale_method == "bislerp":
+            return bislerp(s, width, height)
+        elif upscale_method == "lanczos":
+            return lanczos(s, width, height)
+        else:
+            return torch.nn.functional.interpolate(s, size=(height, width), mode=upscale_method)
+
+def get_tiled_scale_steps(width, height, tile_x, tile_y, overlap):
+    return math.ceil((height / (tile_y - overlap))) * math.ceil((width / (tile_x - overlap)))
+
+@torch.inference_mode()
+def tiled_scale(samples, function, tile_x=64, tile_y=64, overlap = 8, upscale_amount = 4, out_channels = 3, pbar = None):
+    output = torch.empty((samples.shape[0], out_channels, round(samples.shape[2] * upscale_amount), round(samples.shape[3] * upscale_amount)), device="cpu")
+    for b in range(samples.shape[0]):
+        s = samples[b:b+1]
+        out = torch.zeros((s.shape[0], out_channels, round(s.shape[2] * upscale_amount), round(s.shape[3] * upscale_amount)), device="cpu")
+        out_div = torch.zeros((s.shape[0], out_channels, round(s.shape[2] * upscale_amount), round(s.shape[3] * upscale_amount)), device="cpu")
+        for y in range(0, s.shape[2], tile_y - overlap):
+            for x in range(0, s.shape[3], tile_x - overlap):
+                s_in = s[:,:,y:y+tile_y,x:x+tile_x]
+
+                ps = function(s_in).cpu()
+                mask = torch.ones_like(ps)
+                feather = round(overlap * upscale_amount)
+                for t in range(feather):
+                        mask[:,:,t:1+t,:] *= ((1.0/feather) * (t + 1))
+                        mask[:,:,mask.shape[2] -1 -t: mask.shape[2]-t,:] *= ((1.0/feather) * (t + 1))
+                        mask[:,:,:,t:1+t] *= ((1.0/feather) * (t + 1))
+                        mask[:,:,:,mask.shape[3]- 1 - t: mask.shape[3]- t] *= ((1.0/feather) * (t + 1))
+                out[:,:,round(y*upscale_amount):round((y+tile_y)*upscale_amount),round(x*upscale_amount):round((x+tile_x)*upscale_amount)] += ps * mask
+                out_div[:,:,round(y*upscale_amount):round((y+tile_y)*upscale_amount),round(x*upscale_amount):round((x+tile_x)*upscale_amount)] += mask
+                if pbar is not None:
+                    pbar.update(1)
+
+        output[b:b+1] = out/out_div
+    return output
+
+PROGRESS_BAR_ENABLED = True
+def set_progress_bar_enabled(enabled):
+    global PROGRESS_BAR_ENABLED
+    PROGRESS_BAR_ENABLED = enabled
+
+PROGRESS_BAR_HOOK = None
+def set_progress_bar_global_hook(function):
+    global PROGRESS_BAR_HOOK
+    PROGRESS_BAR_HOOK = function
+
+class ProgressBar:
+    def __init__(self, total):
+        global PROGRESS_BAR_HOOK
+        self.total = total
+        self.current = 0
+        self.hook = PROGRESS_BAR_HOOK
+
+    def update_absolute(self, value, total=None, preview=None):
+        if total is not None:
+            self.total = total
+        if value > self.total:
+            value = self.total
+        self.current = value
+        if self.hook is not None:
+            self.hook(self.current, self.total, preview)
+
+    def update(self, value):
+        self.update_absolute(self.current + value)

backend/headless/fcbh_extras/chainner_models/architecture/LICENSE-DAT

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

backend/headless/fcbh_extras/chainner_models/architecture/LICENSE-ESRGAN

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

backend/headless/fcbh_extras/chainner_models/architecture/LICENSE-HAT

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2022 Xiangyu Chen
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

backend/headless/fcbh_extras/chainner_models/architecture/LICENSE-RealESRGAN

@@ -0,0 +1,29 @@
+BSD 3-Clause License
+
+Copyright (c) 2021, Xintao Wang
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+1. Redistributions of source code must retain the above copyright notice, this
+   list of conditions and the following disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright notice,
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+
+3. Neither the name of the copyright holder nor the names of its
+   contributors may be used to endorse or promote products derived from
+   this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

backend/headless/fcbh_extras/chainner_models/architecture/LICENSE-SCUNet

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2022 Kai Zhang (cskaizhang@gmail.com, https://cszn.github.io/). All rights reserved.
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

backend/headless/fcbh_extras/chainner_models/architecture/LICENSE-SPSR

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2018-2022 BasicSR Authors
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

backend/headless/fcbh_extras/chainner_models/architecture/LICENSE-SwiftSRGAN

@@ -0,0 +1,121 @@
+Creative Commons Legal Code
+
+CC0 1.0 Universal
+
+    CREATIVE COMMONS CORPORATION IS NOT A LAW FIRM AND DOES NOT PROVIDE
+    LEGAL SERVICES. DISTRIBUTION OF THIS DOCUMENT DOES NOT CREATE AN
+    ATTORNEY-CLIENT RELATIONSHIP. CREATIVE COMMONS PROVIDES THIS
+    INFORMATION ON AN "AS-IS" BASIS. CREATIVE COMMONS MAKES NO WARRANTIES
+    REGARDING THE USE OF THIS DOCUMENT OR THE INFORMATION OR WORKS
+    PROVIDED HEREUNDER, AND DISCLAIMS LIABILITY FOR DAMAGES RESULTING FROM
+    THE USE OF THIS DOCUMENT OR THE INFORMATION OR WORKS PROVIDED
+    HEREUNDER.
+
+Statement of Purpose
+
+The laws of most jurisdictions throughout the world automatically confer
+exclusive Copyright and Related Rights (defined below) upon the creator
+and subsequent owner(s) (each and all, an "owner") of an original work of
+authorship and/or a database (each, a "Work").
+
+Certain owners wish to permanently relinquish those rights to a Work for
+the purpose of contributing to a commons of creative, cultural and
+scientific works ("Commons") that the public can reliably and without fear
+of later claims of infringement build upon, modify, incorporate in other
+works, reuse and redistribute as freely as possible in any form whatsoever
+and for any purposes, including without limitation commercial purposes.
+These owners may contribute to the Commons to promote the ideal of a free
+culture and the further production of creative, cultural and scientific
+works, or to gain reputation or greater distribution for their Work in
+part through the use and efforts of others.
+
+For these and/or other purposes and motivations, and without any
+expectation of additional consideration or compensation, the person
+associating CC0 with a Work (the "Affirmer"), to the extent that he or she
+is an owner of Copyright and Related Rights in the Work, voluntarily
+elects to apply CC0 to the Work and publicly distribute the Work under its
+terms, with knowledge of his or her Copyright and Related Rights in the
+Work and the meaning and intended legal effect of CC0 on those rights.
+
+1. Copyright and Related Rights. A Work made available under CC0 may be
+protected by copyright and related or neighboring rights ("Copyright and
+Related Rights"). Copyright and Related Rights include, but are not
+limited to, the following:
+
+  i. the right to reproduce, adapt, distribute, perform, display,
+     communicate, and translate a Work;
+ ii. moral rights retained by the original author(s) and/or performer(s);
+iii. publicity and privacy rights pertaining to a person's image or
+     likeness depicted in a Work;
+ iv. rights protecting against unfair competition in regards to a Work,
+     subject to the limitations in paragraph 4(a), below;
+  v. rights protecting the extraction, dissemination, use and reuse of data
+     in a Work;
+ vi. database rights (such as those arising under Directive 96/9/EC of the
+     European Parliament and of the Council of 11 March 1996 on the legal
+     protection of databases, and under any national implementation
+     thereof, including any amended or successor version of such
+     directive); and
+vii. other similar, equivalent or corresponding rights throughout the
+     world based on applicable law or treaty, and any national
+     implementations thereof.
+
+2. Waiver. To the greatest extent permitted by, but not in contravention
+of, applicable law, Affirmer hereby overtly, fully, permanently,
+irrevocably and unconditionally waives, abandons, and surrenders all of
+Affirmer's Copyright and Related Rights and associated claims and causes
+of action, whether now known or unknown (including existing as well as
+future claims and causes of action), in the Work (i) in all territories
+worldwide, (ii) for the maximum duration provided by applicable law or
+treaty (including future time extensions), (iii) in any current or future
+medium and for any number of copies, and (iv) for any purpose whatsoever,
+including without limitation commercial, advertising or promotional
+purposes (the "Waiver"). Affirmer makes the Waiver for the benefit of each
+member of the public at large and to the detriment of Affirmer's heirs and
+successors, fully intending that such Waiver shall not be subject to
+revocation, rescission, cancellation, termination, or any other legal or
+equitable action to disrupt the quiet enjoyment of the Work by the public
+as contemplated by Affirmer's express Statement of Purpose.
+
+3. Public License Fallback. Should any part of the Waiver for any reason
+be judged legally invalid or ineffective under applicable law, then the
+Waiver shall be preserved to the maximum extent permitted taking into
+account Affirmer's express Statement of Purpose. In addition, to the
+extent the Waiver is so judged Affirmer hereby grants to each affected
+person a royalty-free, non transferable, non sublicensable, non exclusive,
+irrevocable and unconditional license to exercise Affirmer's Copyright and
+Related Rights in the Work (i) in all territories worldwide, (ii) for the
+maximum duration provided by applicable law or treaty (including future
+time extensions), (iii) in any current or future medium and for any number
+of copies, and (iv) for any purpose whatsoever, including without
+limitation commercial, advertising or promotional purposes (the
+"License"). The License shall be deemed effective as of the date CC0 was
+applied by Affirmer to the Work. Should any part of the License for any
+reason be judged legally invalid or ineffective under applicable law, such
+partial invalidity or ineffectiveness shall not invalidate the remainder
+of the License, and in such case Affirmer hereby affirms that he or she
+will not (i) exercise any of his or her remaining Copyright and Related
+Rights in the Work or (ii) assert any associated claims and causes of
+action with respect to the Work, in either case contrary to Affirmer's
+express Statement of Purpose.
+
+4. Limitations and Disclaimers.
+
+ a. No trademark or patent rights held by Affirmer are waived, abandoned,
+    surrendered, licensed or otherwise affected by this document.
+ b. Affirmer offers the Work as-is and makes no representations or
+    warranties of any kind concerning the Work, express, implied,
+    statutory or otherwise, including without limitation warranties of
+    title, merchantability, fitness for a particular purpose, non
+    infringement, or the absence of latent or other defects, accuracy, or
+    the present or absence of errors, whether or not discoverable, all to
+    the greatest extent permissible under applicable law.
+ c. Affirmer disclaims responsibility for clearing rights of other persons
+    that may apply to the Work or any use thereof, including without
+    limitation any person's Copyright and Related Rights in the Work.
+    Further, Affirmer disclaims responsibility for obtaining any necessary
+    consents, permissions or other rights required for any use of the
+    Work.
+ d. Affirmer understands and acknowledges that Creative Commons is not a
+    party to this document and has no duty or obligation with respect to
+    this CC0 or use of the Work.

backend/headless/fcbh_extras/chainner_models/architecture/LICENSE-Swin2SR

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [2021] [SwinIR Authors]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

backend/headless/fcbh_extras/chainner_models/architecture/LICENSE-SwinIR

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [2021] [SwinIR Authors]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

backend/headless/fcbh_extras/chainner_models/architecture/LICENSE-lama

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [2021] Samsung Research
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

backend/headless/fcbh_extras/chainner_models/architecture/LaMa.py

@@ -0,0 +1,694 @@
+# pylint: skip-file
+"""
+Model adapted from advimman's lama project: https://github.com/advimman/lama
+"""
+
+# Fast Fourier Convolution NeurIPS 2020
+# original implementation https://github.com/pkumivision/FFC/blob/main/model_zoo/ffc.py
+# paper https://proceedings.neurips.cc/paper/2020/file/2fd5d41ec6cfab47e32164d5624269b1-Paper.pdf
+
+from typing import List
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torchvision.transforms.functional import InterpolationMode, rotate
+
+
+class LearnableSpatialTransformWrapper(nn.Module):
+    def __init__(self, impl, pad_coef=0.5, angle_init_range=80, train_angle=True):
+        super().__init__()
+        self.impl = impl
+        self.angle = torch.rand(1) * angle_init_range
+        if train_angle:
+            self.angle = nn.Parameter(self.angle, requires_grad=True)
+        self.pad_coef = pad_coef
+
+    def forward(self, x):
+        if torch.is_tensor(x):
+            return self.inverse_transform(self.impl(self.transform(x)), x)
+        elif isinstance(x, tuple):
+            x_trans = tuple(self.transform(elem) for elem in x)
+            y_trans = self.impl(x_trans)
+            return tuple(
+                self.inverse_transform(elem, orig_x) for elem, orig_x in zip(y_trans, x)
+            )
+        else:
+            raise ValueError(f"Unexpected input type {type(x)}")
+
+    def transform(self, x):
+        height, width = x.shape[2:]
+        pad_h, pad_w = int(height * self.pad_coef), int(width * self.pad_coef)
+        x_padded = F.pad(x, [pad_w, pad_w, pad_h, pad_h], mode="reflect")
+        x_padded_rotated = rotate(
+            x_padded, self.angle.to(x_padded), InterpolationMode.BILINEAR, fill=0
+        )
+
+        return x_padded_rotated
+
+    def inverse_transform(self, y_padded_rotated, orig_x):
+        height, width = orig_x.shape[2:]
+        pad_h, pad_w = int(height * self.pad_coef), int(width * self.pad_coef)
+
+        y_padded = rotate(
+            y_padded_rotated,
+            -self.angle.to(y_padded_rotated),
+            InterpolationMode.BILINEAR,
+            fill=0,
+        )
+        y_height, y_width = y_padded.shape[2:]
+        y = y_padded[:, :, pad_h : y_height - pad_h, pad_w : y_width - pad_w]
+        return y
+
+
+class SELayer(nn.Module):
+    def __init__(self, channel, reduction=16):
+        super(SELayer, self).__init__()
+        self.avg_pool = nn.AdaptiveAvgPool2d(1)
+        self.fc = nn.Sequential(
+            nn.Linear(channel, channel // reduction, bias=False),
+            nn.ReLU(inplace=True),
+            nn.Linear(channel // reduction, channel, bias=False),
+            nn.Sigmoid(),
+        )
+
+    def forward(self, x):
+        b, c, _, _ = x.size()
+        y = self.avg_pool(x).view(b, c)
+        y = self.fc(y).view(b, c, 1, 1)
+        res = x * y.expand_as(x)
+        return res
+
+
+class FourierUnit(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        out_channels,
+        groups=1,
+        spatial_scale_factor=None,
+        spatial_scale_mode="bilinear",
+        spectral_pos_encoding=False,
+        use_se=False,
+        se_kwargs=None,
+        ffc3d=False,
+        fft_norm="ortho",
+    ):
+        # bn_layer not used
+        super(FourierUnit, self).__init__()
+        self.groups = groups
+
+        self.conv_layer = torch.nn.Conv2d(
+            in_channels=in_channels * 2 + (2 if spectral_pos_encoding else 0),
+            out_channels=out_channels * 2,
+            kernel_size=1,
+            stride=1,
+            padding=0,
+            groups=self.groups,
+            bias=False,
+        )
+        self.bn = torch.nn.BatchNorm2d(out_channels * 2)
+        self.relu = torch.nn.ReLU(inplace=True)
+
+        # squeeze and excitation block
+        self.use_se = use_se
+        if use_se:
+            if se_kwargs is None:
+                se_kwargs = {}
+            self.se = SELayer(self.conv_layer.in_channels, **se_kwargs)
+
+        self.spatial_scale_factor = spatial_scale_factor
+        self.spatial_scale_mode = spatial_scale_mode
+        self.spectral_pos_encoding = spectral_pos_encoding
+        self.ffc3d = ffc3d
+        self.fft_norm = fft_norm
+
+    def forward(self, x):
+        half_check = False
+        if x.type() == "torch.cuda.HalfTensor":
+            # half only works on gpu anyway
+            half_check = True
+
+        batch = x.shape[0]
+
+        if self.spatial_scale_factor is not None:
+            orig_size = x.shape[-2:]
+            x = F.interpolate(
+                x,
+                scale_factor=self.spatial_scale_factor,
+                mode=self.spatial_scale_mode,
+                align_corners=False,
+            )
+
+        # (batch, c, h, w/2+1, 2)
+        fft_dim = (-3, -2, -1) if self.ffc3d else (-2, -1)
+        if half_check == True:
+            ffted = torch.fft.rfftn(
+                x.float(), dim=fft_dim, norm=self.fft_norm
+            )  # .type(torch.cuda.HalfTensor)
+        else:
+            ffted = torch.fft.rfftn(x, dim=fft_dim, norm=self.fft_norm)
+
+        ffted = torch.stack((ffted.real, ffted.imag), dim=-1)
+        ffted = ffted.permute(0, 1, 4, 2, 3).contiguous()  # (batch, c, 2, h, w/2+1)
+        ffted = ffted.view(
+            (
+                batch,
+                -1,
+            )
+            + ffted.size()[3:]
+        )
+
+        if self.spectral_pos_encoding:
+            height, width = ffted.shape[-2:]
+            coords_vert = (
+                torch.linspace(0, 1, height)[None, None, :, None]
+                .expand(batch, 1, height, width)
+                .to(ffted)
+            )
+            coords_hor = (
+                torch.linspace(0, 1, width)[None, None, None, :]
+                .expand(batch, 1, height, width)
+                .to(ffted)
+            )
+            ffted = torch.cat((coords_vert, coords_hor, ffted), dim=1)
+
+        if self.use_se:
+            ffted = self.se(ffted)
+
+        if half_check == True:
+            ffted = self.conv_layer(ffted.half())  # (batch, c*2, h, w/2+1)
+        else:
+            ffted = self.conv_layer(
+                ffted
+            )  # .type(torch.cuda.FloatTensor)  # (batch, c*2, h, w/2+1)
+
+        ffted = self.relu(self.bn(ffted))
+        # forcing to be always float
+        ffted = ffted.float()
+
+        ffted = (
+            ffted.view(
+                (
+                    batch,
+                    -1,
+                    2,
+                )
+                + ffted.size()[2:]
+            )
+            .permute(0, 1, 3, 4, 2)
+            .contiguous()
+        )  # (batch,c, t, h, w/2+1, 2)
+
+        ffted = torch.complex(ffted[..., 0], ffted[..., 1])
+
+        ifft_shape_slice = x.shape[-3:] if self.ffc3d else x.shape[-2:]
+        output = torch.fft.irfftn(
+            ffted, s=ifft_shape_slice, dim=fft_dim, norm=self.fft_norm
+        )
+
+        if half_check == True:
+            output = output.half()
+
+        if self.spatial_scale_factor is not None:
+            output = F.interpolate(
+                output,
+                size=orig_size,
+                mode=self.spatial_scale_mode,
+                align_corners=False,
+            )
+
+        return output
+
+
+class SpectralTransform(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        out_channels,
+        stride=1,
+        groups=1,
+        enable_lfu=True,
+        separable_fu=False,
+        **fu_kwargs,
+    ):
+        # bn_layer not used
+        super(SpectralTransform, self).__init__()
+        self.enable_lfu = enable_lfu
+        if stride == 2:
+            self.downsample = nn.AvgPool2d(kernel_size=(2, 2), stride=2)
+        else:
+            self.downsample = nn.Identity()
+
+        self.stride = stride
+        self.conv1 = nn.Sequential(
+            nn.Conv2d(
+                in_channels, out_channels // 2, kernel_size=1, groups=groups, bias=False
+            ),
+            nn.BatchNorm2d(out_channels // 2),
+            nn.ReLU(inplace=True),
+        )
+        fu_class = FourierUnit
+        self.fu = fu_class(out_channels // 2, out_channels // 2, groups, **fu_kwargs)
+        if self.enable_lfu:
+            self.lfu = fu_class(out_channels // 2, out_channels // 2, groups)
+        self.conv2 = torch.nn.Conv2d(
+            out_channels // 2, out_channels, kernel_size=1, groups=groups, bias=False
+        )
+
+    def forward(self, x):
+        x = self.downsample(x)
+        x = self.conv1(x)
+        output = self.fu(x)
+
+        if self.enable_lfu:
+            _, c, h, _ = x.shape
+            split_no = 2
+            split_s = h // split_no
+            xs = torch.cat(
+                torch.split(x[:, : c // 4], split_s, dim=-2), dim=1
+            ).contiguous()
+            xs = torch.cat(torch.split(xs, split_s, dim=-1), dim=1).contiguous()
+            xs = self.lfu(xs)
+            xs = xs.repeat(1, 1, split_no, split_no).contiguous()
+        else:
+            xs = 0
+
+        output = self.conv2(x + output + xs)
+
+        return output
+
+
+class FFC(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        out_channels,
+        kernel_size,
+        ratio_gin,
+        ratio_gout,
+        stride=1,
+        padding=0,
+        dilation=1,
+        groups=1,
+        bias=False,
+        enable_lfu=True,
+        padding_type="reflect",
+        gated=False,
+        **spectral_kwargs,
+    ):
+        super(FFC, self).__init__()
+
+        assert stride == 1 or stride == 2, "Stride should be 1 or 2."
+        self.stride = stride
+
+        in_cg = int(in_channels * ratio_gin)
+        in_cl = in_channels - in_cg
+        out_cg = int(out_channels * ratio_gout)
+        out_cl = out_channels - out_cg
+        # groups_g = 1 if groups == 1 else int(groups * ratio_gout)
+        # groups_l = 1 if groups == 1 else groups - groups_g
+
+        self.ratio_gin = ratio_gin
+        self.ratio_gout = ratio_gout
+        self.global_in_num = in_cg
+
+        module = nn.Identity if in_cl == 0 or out_cl == 0 else nn.Conv2d
+        self.convl2l = module(
+            in_cl,
+            out_cl,
+            kernel_size,
+            stride,
+            padding,
+            dilation,
+            groups,
+            bias,
+            padding_mode=padding_type,
+        )
+        module = nn.Identity if in_cl == 0 or out_cg == 0 else nn.Conv2d
+        self.convl2g = module(
+            in_cl,
+            out_cg,
+            kernel_size,
+            stride,
+            padding,
+            dilation,
+            groups,
+            bias,
+            padding_mode=padding_type,
+        )
+        module = nn.Identity if in_cg == 0 or out_cl == 0 else nn.Conv2d
+        self.convg2l = module(
+            in_cg,
+            out_cl,
+            kernel_size,
+            stride,
+            padding,
+            dilation,
+            groups,
+            bias,
+            padding_mode=padding_type,
+        )
+        module = nn.Identity if in_cg == 0 or out_cg == 0 else SpectralTransform
+        self.convg2g = module(
+            in_cg,
+            out_cg,
+            stride,
+            1 if groups == 1 else groups // 2,
+            enable_lfu,
+            **spectral_kwargs,
+        )
+
+        self.gated = gated
+        module = (
+            nn.Identity if in_cg == 0 or out_cl == 0 or not self.gated else nn.Conv2d
+        )
+        self.gate = module(in_channels, 2, 1)
+
+    def forward(self, x):
+        x_l, x_g = x if type(x) is tuple else (x, 0)
+        out_xl, out_xg = 0, 0
+
+        if self.gated:
+            total_input_parts = [x_l]
+            if torch.is_tensor(x_g):
+                total_input_parts.append(x_g)
+            total_input = torch.cat(total_input_parts, dim=1)
+
+            gates = torch.sigmoid(self.gate(total_input))
+            g2l_gate, l2g_gate = gates.chunk(2, dim=1)
+        else:
+            g2l_gate, l2g_gate = 1, 1
+
+        if self.ratio_gout != 1:
+            out_xl = self.convl2l(x_l) + self.convg2l(x_g) * g2l_gate
+        if self.ratio_gout != 0:
+            out_xg = self.convl2g(x_l) * l2g_gate + self.convg2g(x_g)
+
+        return out_xl, out_xg
+
+
+class FFC_BN_ACT(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        out_channels,
+        kernel_size,
+        ratio_gin,
+        ratio_gout,
+        stride=1,
+        padding=0,
+        dilation=1,
+        groups=1,
+        bias=False,
+        norm_layer=nn.BatchNorm2d,
+        activation_layer=nn.Identity,
+        padding_type="reflect",
+        enable_lfu=True,
+        **kwargs,
+    ):
+        super(FFC_BN_ACT, self).__init__()
+        self.ffc = FFC(
+            in_channels,
+            out_channels,
+            kernel_size,
+            ratio_gin,
+            ratio_gout,
+            stride,
+            padding,
+            dilation,
+            groups,
+            bias,
+            enable_lfu,
+            padding_type=padding_type,
+            **kwargs,
+        )
+        lnorm = nn.Identity if ratio_gout == 1 else norm_layer
+        gnorm = nn.Identity if ratio_gout == 0 else norm_layer
+        global_channels = int(out_channels * ratio_gout)
+        self.bn_l = lnorm(out_channels - global_channels)
+        self.bn_g = gnorm(global_channels)
+
+        lact = nn.Identity if ratio_gout == 1 else activation_layer
+        gact = nn.Identity if ratio_gout == 0 else activation_layer
+        self.act_l = lact(inplace=True)
+        self.act_g = gact(inplace=True)
+
+    def forward(self, x):
+        x_l, x_g = self.ffc(x)
+        x_l = self.act_l(self.bn_l(x_l))
+        x_g = self.act_g(self.bn_g(x_g))
+        return x_l, x_g
+
+
+class FFCResnetBlock(nn.Module):
+    def __init__(
+        self,
+        dim,
+        padding_type,
+        norm_layer,
+        activation_layer=nn.ReLU,
+        dilation=1,
+        spatial_transform_kwargs=None,
+        inline=False,
+        **conv_kwargs,
+    ):
+        super().__init__()
+        self.conv1 = FFC_BN_ACT(
+            dim,
+            dim,
+            kernel_size=3,
+            padding=dilation,
+            dilation=dilation,
+            norm_layer=norm_layer,
+            activation_layer=activation_layer,
+            padding_type=padding_type,
+            **conv_kwargs,
+        )
+        self.conv2 = FFC_BN_ACT(
+            dim,
+            dim,
+            kernel_size=3,
+            padding=dilation,
+            dilation=dilation,
+            norm_layer=norm_layer,
+            activation_layer=activation_layer,
+            padding_type=padding_type,
+            **conv_kwargs,
+        )
+        if spatial_transform_kwargs is not None:
+            self.conv1 = LearnableSpatialTransformWrapper(
+                self.conv1, **spatial_transform_kwargs
+            )
+            self.conv2 = LearnableSpatialTransformWrapper(
+                self.conv2, **spatial_transform_kwargs
+            )
+        self.inline = inline
+
+    def forward(self, x):
+        if self.inline:
+            x_l, x_g = (
+                x[:, : -self.conv1.ffc.global_in_num],
+                x[:, -self.conv1.ffc.global_in_num :],
+            )
+        else:
+            x_l, x_g = x if type(x) is tuple else (x, 0)
+
+        id_l, id_g = x_l, x_g
+
+        x_l, x_g = self.conv1((x_l, x_g))
+        x_l, x_g = self.conv2((x_l, x_g))
+
+        x_l, x_g = id_l + x_l, id_g + x_g
+        out = x_l, x_g
+        if self.inline:
+            out = torch.cat(out, dim=1)
+        return out
+
+
+class ConcatTupleLayer(nn.Module):
+    def forward(self, x):
+        assert isinstance(x, tuple)
+        x_l, x_g = x
+        assert torch.is_tensor(x_l) or torch.is_tensor(x_g)
+        if not torch.is_tensor(x_g):
+            return x_l
+        return torch.cat(x, dim=1)
+
+
+class FFCResNetGenerator(nn.Module):
+    def __init__(
+        self,
+        input_nc,
+        output_nc,
+        ngf=64,
+        n_downsampling=3,
+        n_blocks=18,
+        norm_layer=nn.BatchNorm2d,
+        padding_type="reflect",
+        activation_layer=nn.ReLU,
+        up_norm_layer=nn.BatchNorm2d,
+        up_activation=nn.ReLU(True),
+        init_conv_kwargs={},
+        downsample_conv_kwargs={},
+        resnet_conv_kwargs={},
+        spatial_transform_layers=None,
+        spatial_transform_kwargs={},
+        max_features=1024,
+        out_ffc=False,
+        out_ffc_kwargs={},
+    ):
+        assert n_blocks >= 0
+        super().__init__()
+        """
+        init_conv_kwargs = {'ratio_gin': 0, 'ratio_gout': 0, 'enable_lfu': False}
+        downsample_conv_kwargs = {'ratio_gin': '${generator.init_conv_kwargs.ratio_gout}', 'ratio_gout': '${generator.downsample_conv_kwargs.ratio_gin}', 'enable_lfu': False}
+        resnet_conv_kwargs = {'ratio_gin': 0.75, 'ratio_gout': '${generator.resnet_conv_kwargs.ratio_gin}', 'enable_lfu': False}
+        spatial_transform_kwargs = {}
+        out_ffc_kwargs = {}
+        """
+        """
+        print(input_nc, output_nc, ngf, n_downsampling, n_blocks, norm_layer,
+                padding_type, activation_layer,
+                up_norm_layer, up_activation,
+                spatial_transform_layers,
+                add_out_act, max_features, out_ffc, file=sys.stderr)
+
+        4 3 64 3 18 <class 'torch.nn.modules.batchnorm.BatchNorm2d'>
+        reflect <class 'torch.nn.modules.activation.ReLU'>
+        <class 'torch.nn.modules.batchnorm.BatchNorm2d'>
+        ReLU(inplace=True)
+        None sigmoid 1024 False
+        """
+        init_conv_kwargs = {"ratio_gin": 0, "ratio_gout": 0, "enable_lfu": False}
+        downsample_conv_kwargs = {"ratio_gin": 0, "ratio_gout": 0, "enable_lfu": False}
+        resnet_conv_kwargs = {
+            "ratio_gin": 0.75,
+            "ratio_gout": 0.75,
+            "enable_lfu": False,
+        }
+        spatial_transform_kwargs = {}
+        out_ffc_kwargs = {}
+
+        model = [
+            nn.ReflectionPad2d(3),
+            FFC_BN_ACT(
+                input_nc,
+                ngf,
+                kernel_size=7,
+                padding=0,
+                norm_layer=norm_layer,
+                activation_layer=activation_layer,
+                **init_conv_kwargs,
+            ),
+        ]
+
+        ### downsample
+        for i in range(n_downsampling):
+            mult = 2**i
+            if i == n_downsampling - 1:
+                cur_conv_kwargs = dict(downsample_conv_kwargs)
+                cur_conv_kwargs["ratio_gout"] = resnet_conv_kwargs.get("ratio_gin", 0)
+            else:
+                cur_conv_kwargs = downsample_conv_kwargs
+            model += [
+                FFC_BN_ACT(
+                    min(max_features, ngf * mult),
+                    min(max_features, ngf * mult * 2),
+                    kernel_size=3,
+                    stride=2,
+                    padding=1,
+                    norm_layer=norm_layer,
+                    activation_layer=activation_layer,
+                    **cur_conv_kwargs,
+                )
+            ]
+
+        mult = 2**n_downsampling
+        feats_num_bottleneck = min(max_features, ngf * mult)
+
+        ### resnet blocks
+        for i in range(n_blocks):
+            cur_resblock = FFCResnetBlock(
+                feats_num_bottleneck,
+                padding_type=padding_type,
+                activation_layer=activation_layer,
+                norm_layer=norm_layer,
+                **resnet_conv_kwargs,
+            )
+            if spatial_transform_layers is not None and i in spatial_transform_layers:
+                cur_resblock = LearnableSpatialTransformWrapper(
+                    cur_resblock, **spatial_transform_kwargs
+                )
+            model += [cur_resblock]
+
+        model += [ConcatTupleLayer()]
+
+        ### upsample
+        for i in range(n_downsampling):
+            mult = 2 ** (n_downsampling - i)
+            model += [
+                nn.ConvTranspose2d(
+                    min(max_features, ngf * mult),
+                    min(max_features, int(ngf * mult / 2)),
+                    kernel_size=3,
+                    stride=2,
+                    padding=1,
+                    output_padding=1,
+                ),
+                up_norm_layer(min(max_features, int(ngf * mult / 2))),
+                up_activation,
+            ]
+
+        if out_ffc:
+            model += [
+                FFCResnetBlock(
+                    ngf,
+                    padding_type=padding_type,
+                    activation_layer=activation_layer,
+                    norm_layer=norm_layer,
+                    inline=True,
+                    **out_ffc_kwargs,
+                )
+            ]
+
+        model += [
+            nn.ReflectionPad2d(3),
+            nn.Conv2d(ngf, output_nc, kernel_size=7, padding=0),
+        ]
+        model.append(nn.Sigmoid())
+        self.model = nn.Sequential(*model)
+
+    def forward(self, image, mask):
+        return self.model(torch.cat([image, mask], dim=1))
+
+
+class LaMa(nn.Module):
+    def __init__(self, state_dict) -> None:
+        super(LaMa, self).__init__()
+        self.model_arch = "LaMa"
+        self.sub_type = "Inpaint"
+        self.in_nc = 4
+        self.out_nc = 3
+        self.scale = 1
+
+        self.min_size = None
+        self.pad_mod = 8
+        self.pad_to_square = False
+
+        self.model = FFCResNetGenerator(self.in_nc, self.out_nc)
+        self.state = {
+            k.replace("generator.model", "model.model"): v
+            for k, v in state_dict.items()
+        }
+
+        self.supports_fp16 = False
+        self.support_bf16 = True
+
+        self.load_state_dict(self.state, strict=False)
+
+    def forward(self, img, mask):
+        masked_img = img * (1 - mask)
+        inpainted_mask = mask * self.model.forward(masked_img, mask)
+        result = inpainted_mask + (1 - mask) * img
+        return result

backend/headless/fcbh_extras/chainner_models/architecture/OmniSR/ChannelAttention.py

@@ -0,0 +1,110 @@
+import math
+
+import torch.nn as nn
+
+
+class CA_layer(nn.Module):
+    def __init__(self, channel, reduction=16):
+        super(CA_layer, self).__init__()
+        # global average pooling
+        self.gap = nn.AdaptiveAvgPool2d(1)
+        self.fc = nn.Sequential(
+            nn.Conv2d(channel, channel // reduction, kernel_size=(1, 1), bias=False),
+            nn.GELU(),
+            nn.Conv2d(channel // reduction, channel, kernel_size=(1, 1), bias=False),
+            # nn.Sigmoid()
+        )
+
+    def forward(self, x):
+        y = self.fc(self.gap(x))
+        return x * y.expand_as(x)
+
+
+class Simple_CA_layer(nn.Module):
+    def __init__(self, channel):
+        super(Simple_CA_layer, self).__init__()
+        self.gap = nn.AdaptiveAvgPool2d(1)
+        self.fc = nn.Conv2d(
+            in_channels=channel,
+            out_channels=channel,
+            kernel_size=1,
+            padding=0,
+            stride=1,
+            groups=1,
+            bias=True,
+        )
+
+    def forward(self, x):
+        return x * self.fc(self.gap(x))
+
+
+class ECA_layer(nn.Module):
+    """Constructs a ECA module.
+    Args:
+        channel: Number of channels of the input feature map
+        k_size: Adaptive selection of kernel size
+    """
+
+    def __init__(self, channel):
+        super(ECA_layer, self).__init__()
+
+        b = 1
+        gamma = 2
+        k_size = int(abs(math.log(channel, 2) + b) / gamma)
+        k_size = k_size if k_size % 2 else k_size + 1
+        self.avg_pool = nn.AdaptiveAvgPool2d(1)
+        self.conv = nn.Conv1d(
+            1, 1, kernel_size=k_size, padding=(k_size - 1) // 2, bias=False
+        )
+        # self.sigmoid = nn.Sigmoid()
+
+    def forward(self, x):
+        # x: input features with shape [b, c, h, w]
+        # b, c, h, w = x.size()
+
+        # feature descriptor on the global spatial information
+        y = self.avg_pool(x)
+
+        # Two different branches of ECA module
+        y = self.conv(y.squeeze(-1).transpose(-1, -2)).transpose(-1, -2).unsqueeze(-1)
+
+        # Multi-scale information fusion
+        # y = self.sigmoid(y)
+
+        return x * y.expand_as(x)
+
+
+class ECA_MaxPool_layer(nn.Module):
+    """Constructs a ECA module.
+    Args:
+        channel: Number of channels of the input feature map
+        k_size: Adaptive selection of kernel size
+    """
+
+    def __init__(self, channel):
+        super(ECA_MaxPool_layer, self).__init__()
+
+        b = 1
+        gamma = 2
+        k_size = int(abs(math.log(channel, 2) + b) / gamma)
+        k_size = k_size if k_size % 2 else k_size + 1
+        self.max_pool = nn.AdaptiveMaxPool2d(1)
+        self.conv = nn.Conv1d(
+            1, 1, kernel_size=k_size, padding=(k_size - 1) // 2, bias=False
+        )
+        # self.sigmoid = nn.Sigmoid()
+
+    def forward(self, x):
+        # x: input features with shape [b, c, h, w]
+        # b, c, h, w = x.size()
+
+        # feature descriptor on the global spatial information
+        y = self.max_pool(x)
+
+        # Two different branches of ECA module
+        y = self.conv(y.squeeze(-1).transpose(-1, -2)).transpose(-1, -2).unsqueeze(-1)
+
+        # Multi-scale information fusion
+        # y = self.sigmoid(y)
+
+        return x * y.expand_as(x)

backend/headless/fcbh_extras/chainner_models/architecture/OmniSR/LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

backend/headless/fcbh_extras/chainner_models/architecture/OmniSR/OSA.py

@@ -0,0 +1,577 @@
+#!/usr/bin/env python3
+# -*- coding:utf-8 -*-
+#############################################################
+# File: OSA.py
+# Created Date: Tuesday April 28th 2022
+# Author: Chen Xuanhong
+# Email: chenxuanhongzju@outlook.com
+# Last Modified:  Sunday, 23rd April 2023 3:07:42 pm
+# Modified By: Chen Xuanhong
+# Copyright (c) 2020 Shanghai Jiao Tong University
+#############################################################
+
+import torch
+import torch.nn.functional as F
+from einops import rearrange, repeat
+from einops.layers.torch import Rearrange, Reduce
+from torch import einsum, nn
+
+from .layernorm import LayerNorm2d
+
+# helpers
+
+
+def exists(val):
+    return val is not None
+
+
+def default(val, d):
+    return val if exists(val) else d
+
+
+def cast_tuple(val, length=1):
+    return val if isinstance(val, tuple) else ((val,) * length)
+
+
+# helper classes
+
+
+class PreNormResidual(nn.Module):
+    def __init__(self, dim, fn):
+        super().__init__()
+        self.norm = nn.LayerNorm(dim)
+        self.fn = fn
+
+    def forward(self, x):
+        return self.fn(self.norm(x)) + x
+
+
+class Conv_PreNormResidual(nn.Module):
+    def __init__(self, dim, fn):
+        super().__init__()
+        self.norm = LayerNorm2d(dim)
+        self.fn = fn
+
+    def forward(self, x):
+        return self.fn(self.norm(x)) + x
+
+
+class FeedForward(nn.Module):
+    def __init__(self, dim, mult=2, dropout=0.0):
+        super().__init__()
+        inner_dim = int(dim * mult)
+        self.net = nn.Sequential(
+            nn.Linear(dim, inner_dim),
+            nn.GELU(),
+            nn.Dropout(dropout),
+            nn.Linear(inner_dim, dim),
+            nn.Dropout(dropout),
+        )
+
+    def forward(self, x):
+        return self.net(x)
+
+
+class Conv_FeedForward(nn.Module):
+    def __init__(self, dim, mult=2, dropout=0.0):
+        super().__init__()
+        inner_dim = int(dim * mult)
+        self.net = nn.Sequential(
+            nn.Conv2d(dim, inner_dim, 1, 1, 0),
+            nn.GELU(),
+            nn.Dropout(dropout),
+            nn.Conv2d(inner_dim, dim, 1, 1, 0),
+            nn.Dropout(dropout),
+        )
+
+    def forward(self, x):
+        return self.net(x)
+
+
+class Gated_Conv_FeedForward(nn.Module):
+    def __init__(self, dim, mult=1, bias=False, dropout=0.0):
+        super().__init__()
+
+        hidden_features = int(dim * mult)
+
+        self.project_in = nn.Conv2d(dim, hidden_features * 2, kernel_size=1, bias=bias)
+
+        self.dwconv = nn.Conv2d(
+            hidden_features * 2,
+            hidden_features * 2,
+            kernel_size=3,
+            stride=1,
+            padding=1,
+            groups=hidden_features * 2,
+            bias=bias,
+        )
+
+        self.project_out = nn.Conv2d(hidden_features, dim, kernel_size=1, bias=bias)
+
+    def forward(self, x):
+        x = self.project_in(x)
+        x1, x2 = self.dwconv(x).chunk(2, dim=1)
+        x = F.gelu(x1) * x2
+        x = self.project_out(x)
+        return x
+
+
+# MBConv
+
+
+class SqueezeExcitation(nn.Module):
+    def __init__(self, dim, shrinkage_rate=0.25):
+        super().__init__()
+        hidden_dim = int(dim * shrinkage_rate)
+
+        self.gate = nn.Sequential(
+            Reduce("b c h w -> b c", "mean"),
+            nn.Linear(dim, hidden_dim, bias=False),
+            nn.SiLU(),
+            nn.Linear(hidden_dim, dim, bias=False),
+            nn.Sigmoid(),
+            Rearrange("b c -> b c 1 1"),
+        )
+
+    def forward(self, x):
+        return x * self.gate(x)
+
+
+class MBConvResidual(nn.Module):
+    def __init__(self, fn, dropout=0.0):
+        super().__init__()
+        self.fn = fn
+        self.dropsample = Dropsample(dropout)
+
+    def forward(self, x):
+        out = self.fn(x)
+        out = self.dropsample(out)
+        return out + x
+
+
+class Dropsample(nn.Module):
+    def __init__(self, prob=0):
+        super().__init__()
+        self.prob = prob
+
+    def forward(self, x):
+        device = x.device
+
+        if self.prob == 0.0 or (not self.training):
+            return x
+
+        keep_mask = (
+            torch.FloatTensor((x.shape[0], 1, 1, 1), device=device).uniform_()
+            > self.prob
+        )
+        return x * keep_mask / (1 - self.prob)
+
+
+def MBConv(
+    dim_in, dim_out, *, downsample, expansion_rate=4, shrinkage_rate=0.25, dropout=0.0
+):
+    hidden_dim = int(expansion_rate * dim_out)
+    stride = 2 if downsample else 1
+
+    net = nn.Sequential(
+        nn.Conv2d(dim_in, hidden_dim, 1),
+        # nn.BatchNorm2d(hidden_dim),
+        nn.GELU(),
+        nn.Conv2d(
+            hidden_dim, hidden_dim, 3, stride=stride, padding=1, groups=hidden_dim
+        ),
+        # nn.BatchNorm2d(hidden_dim),
+        nn.GELU(),
+        SqueezeExcitation(hidden_dim, shrinkage_rate=shrinkage_rate),
+        nn.Conv2d(hidden_dim, dim_out, 1),
+        # nn.BatchNorm2d(dim_out)
+    )
+
+    if dim_in == dim_out and not downsample:
+        net = MBConvResidual(net, dropout=dropout)
+
+    return net
+
+
+# attention related classes
+class Attention(nn.Module):
+    def __init__(
+        self,
+        dim,
+        dim_head=32,
+        dropout=0.0,
+        window_size=7,
+        with_pe=True,
+    ):
+        super().__init__()
+        assert (
+            dim % dim_head
+        ) == 0, "dimension should be divisible by dimension per head"
+
+        self.heads = dim // dim_head
+        self.scale = dim_head**-0.5
+        self.with_pe = with_pe
+
+        self.to_qkv = nn.Linear(dim, dim * 3, bias=False)
+
+        self.attend = nn.Sequential(nn.Softmax(dim=-1), nn.Dropout(dropout))
+
+        self.to_out = nn.Sequential(
+            nn.Linear(dim, dim, bias=False), nn.Dropout(dropout)
+        )
+
+        # relative positional bias
+        if self.with_pe:
+            self.rel_pos_bias = nn.Embedding((2 * window_size - 1) ** 2, self.heads)
+
+            pos = torch.arange(window_size)
+            grid = torch.stack(torch.meshgrid(pos, pos))
+            grid = rearrange(grid, "c i j -> (i j) c")
+            rel_pos = rearrange(grid, "i ... -> i 1 ...") - rearrange(
+                grid, "j ... -> 1 j ..."
+            )
+            rel_pos += window_size - 1
+            rel_pos_indices = (rel_pos * torch.tensor([2 * window_size - 1, 1])).sum(
+                dim=-1
+            )
+
+            self.register_buffer("rel_pos_indices", rel_pos_indices, persistent=False)
+
+    def forward(self, x):
+        batch, height, width, window_height, window_width, _, device, h = (
+            *x.shape,
+            x.device,
+            self.heads,
+        )
+
+        # flatten
+
+        x = rearrange(x, "b x y w1 w2 d -> (b x y) (w1 w2) d")
+
+        # project for queries, keys, values
+
+        q, k, v = self.to_qkv(x).chunk(3, dim=-1)
+
+        # split heads
+
+        q, k, v = map(lambda t: rearrange(t, "b n (h d ) -> b h n d", h=h), (q, k, v))
+
+        # scale
+
+        q = q * self.scale
+
+        # sim
+
+        sim = einsum("b h i d, b h j d -> b h i j", q, k)
+
+        # add positional bias
+        if self.with_pe:
+            bias = self.rel_pos_bias(self.rel_pos_indices)
+            sim = sim + rearrange(bias, "i j h -> h i j")
+
+        # attention
+
+        attn = self.attend(sim)
+
+        # aggregate
+
+        out = einsum("b h i j, b h j d -> b h i d", attn, v)
+
+        # merge heads
+
+        out = rearrange(
+            out, "b h (w1 w2) d -> b w1 w2 (h d)", w1=window_height, w2=window_width
+        )
+
+        # combine heads out
+
+        out = self.to_out(out)
+        return rearrange(out, "(b x y) ... -> b x y ...", x=height, y=width)
+
+
+class Block_Attention(nn.Module):
+    def __init__(
+        self,
+        dim,
+        dim_head=32,
+        bias=False,
+        dropout=0.0,
+        window_size=7,
+        with_pe=True,
+    ):
+        super().__init__()
+        assert (
+            dim % dim_head
+        ) == 0, "dimension should be divisible by dimension per head"
+
+        self.heads = dim // dim_head
+        self.ps = window_size
+        self.scale = dim_head**-0.5
+        self.with_pe = with_pe
+
+        self.qkv = nn.Conv2d(dim, dim * 3, kernel_size=1, bias=bias)
+        self.qkv_dwconv = nn.Conv2d(
+            dim * 3,
+            dim * 3,
+            kernel_size=3,
+            stride=1,
+            padding=1,
+            groups=dim * 3,
+            bias=bias,
+        )
+
+        self.attend = nn.Sequential(nn.Softmax(dim=-1), nn.Dropout(dropout))
+
+        self.to_out = nn.Conv2d(dim, dim, kernel_size=1, bias=bias)
+
+    def forward(self, x):
+        # project for queries, keys, values
+        b, c, h, w = x.shape
+
+        qkv = self.qkv_dwconv(self.qkv(x))
+        q, k, v = qkv.chunk(3, dim=1)
+
+        # split heads
+
+        q, k, v = map(
+            lambda t: rearrange(
+                t,
+                "b (h d) (x w1) (y w2) -> (b x y) h (w1 w2) d",
+                h=self.heads,
+                w1=self.ps,
+                w2=self.ps,
+            ),
+            (q, k, v),
+        )
+
+        # scale
+
+        q = q * self.scale
+
+        # sim
+
+        sim = einsum("b h i d, b h j d -> b h i j", q, k)
+
+        # attention
+        attn = self.attend(sim)
+
+        # aggregate
+
+        out = einsum("b h i j, b h j d -> b h i d", attn, v)
+
+        # merge heads
+        out = rearrange(
+            out,
+            "(b x y) head (w1 w2) d -> b (head d) (x w1) (y w2)",
+            x=h // self.ps,
+            y=w // self.ps,
+            head=self.heads,
+            w1=self.ps,
+            w2=self.ps,
+        )
+
+        out = self.to_out(out)
+        return out
+
+
+class Channel_Attention(nn.Module):
+    def __init__(self, dim, heads, bias=False, dropout=0.0, window_size=7):
+        super(Channel_Attention, self).__init__()
+        self.heads = heads
+
+        self.temperature = nn.Parameter(torch.ones(heads, 1, 1))
+
+        self.ps = window_size
+
+        self.qkv = nn.Conv2d(dim, dim * 3, kernel_size=1, bias=bias)
+        self.qkv_dwconv = nn.Conv2d(
+            dim * 3,
+            dim * 3,
+            kernel_size=3,
+            stride=1,
+            padding=1,
+            groups=dim * 3,
+            bias=bias,
+        )
+        self.project_out = nn.Conv2d(dim, dim, kernel_size=1, bias=bias)
+
+    def forward(self, x):
+        b, c, h, w = x.shape
+
+        qkv = self.qkv_dwconv(self.qkv(x))
+        qkv = qkv.chunk(3, dim=1)
+
+        q, k, v = map(
+            lambda t: rearrange(
+                t,
+                "b (head d) (h ph) (w pw) -> b (h w) head d (ph pw)",
+                ph=self.ps,
+                pw=self.ps,
+                head=self.heads,
+            ),
+            qkv,
+        )
+
+        q = F.normalize(q, dim=-1)
+        k = F.normalize(k, dim=-1)
+
+        attn = (q @ k.transpose(-2, -1)) * self.temperature
+        attn = attn.softmax(dim=-1)
+        out = attn @ v
+
+        out = rearrange(
+            out,
+            "b (h w) head d (ph pw) -> b (head d) (h ph) (w pw)",
+            h=h // self.ps,
+            w=w // self.ps,
+            ph=self.ps,
+            pw=self.ps,
+            head=self.heads,
+        )
+
+        out = self.project_out(out)
+
+        return out
+
+
+class Channel_Attention_grid(nn.Module):
+    def __init__(self, dim, heads, bias=False, dropout=0.0, window_size=7):
+        super(Channel_Attention_grid, self).__init__()
+        self.heads = heads
+
+        self.temperature = nn.Parameter(torch.ones(heads, 1, 1))
+
+        self.ps = window_size
+
+        self.qkv = nn.Conv2d(dim, dim * 3, kernel_size=1, bias=bias)
+        self.qkv_dwconv = nn.Conv2d(
+            dim * 3,
+            dim * 3,
+            kernel_size=3,
+            stride=1,
+            padding=1,
+            groups=dim * 3,
+            bias=bias,
+        )
+        self.project_out = nn.Conv2d(dim, dim, kernel_size=1, bias=bias)
+
+    def forward(self, x):
+        b, c, h, w = x.shape
+
+        qkv = self.qkv_dwconv(self.qkv(x))
+        qkv = qkv.chunk(3, dim=1)
+
+        q, k, v = map(
+            lambda t: rearrange(
+                t,
+                "b (head d) (h ph) (w pw) -> b (ph pw) head d (h w)",
+                ph=self.ps,
+                pw=self.ps,
+                head=self.heads,
+            ),
+            qkv,
+        )
+
+        q = F.normalize(q, dim=-1)
+        k = F.normalize(k, dim=-1)
+
+        attn = (q @ k.transpose(-2, -1)) * self.temperature
+        attn = attn.softmax(dim=-1)
+        out = attn @ v
+
+        out = rearrange(
+            out,
+            "b (ph pw) head d (h w) -> b (head d) (h ph) (w pw)",
+            h=h // self.ps,
+            w=w // self.ps,
+            ph=self.ps,
+            pw=self.ps,
+            head=self.heads,
+        )
+
+        out = self.project_out(out)
+
+        return out
+
+
+class OSA_Block(nn.Module):
+    def __init__(
+        self,
+        channel_num=64,
+        bias=True,
+        ffn_bias=True,
+        window_size=8,
+        with_pe=False,
+        dropout=0.0,
+    ):
+        super(OSA_Block, self).__init__()
+
+        w = window_size
+
+        self.layer = nn.Sequential(
+            MBConv(
+                channel_num,
+                channel_num,
+                downsample=False,
+                expansion_rate=1,
+                shrinkage_rate=0.25,
+            ),
+            Rearrange(
+                "b d (x w1) (y w2) -> b x y w1 w2 d", w1=w, w2=w
+            ),  # block-like attention
+            PreNormResidual(
+                channel_num,
+                Attention(
+                    dim=channel_num,
+                    dim_head=channel_num // 4,
+                    dropout=dropout,
+                    window_size=window_size,
+                    with_pe=with_pe,
+                ),
+            ),
+            Rearrange("b x y w1 w2 d -> b d (x w1) (y w2)"),
+            Conv_PreNormResidual(
+                channel_num, Gated_Conv_FeedForward(dim=channel_num, dropout=dropout)
+            ),
+            # channel-like attention
+            Conv_PreNormResidual(
+                channel_num,
+                Channel_Attention(
+                    dim=channel_num, heads=4, dropout=dropout, window_size=window_size
+                ),
+            ),
+            Conv_PreNormResidual(
+                channel_num, Gated_Conv_FeedForward(dim=channel_num, dropout=dropout)
+            ),
+            Rearrange(
+                "b d (w1 x) (w2 y) -> b x y w1 w2 d", w1=w, w2=w
+            ),  # grid-like attention
+            PreNormResidual(
+                channel_num,
+                Attention(
+                    dim=channel_num,
+                    dim_head=channel_num // 4,
+                    dropout=dropout,
+                    window_size=window_size,
+                    with_pe=with_pe,
+                ),
+            ),
+            Rearrange("b x y w1 w2 d -> b d (w1 x) (w2 y)"),
+            Conv_PreNormResidual(
+                channel_num, Gated_Conv_FeedForward(dim=channel_num, dropout=dropout)
+            ),
+            # channel-like attention
+            Conv_PreNormResidual(
+                channel_num,
+                Channel_Attention_grid(
+                    dim=channel_num, heads=4, dropout=dropout, window_size=window_size
+                ),
+            ),
+            Conv_PreNormResidual(
+                channel_num, Gated_Conv_FeedForward(dim=channel_num, dropout=dropout)
+            ),
+        )
+
+    def forward(self, x):
+        out = self.layer(x)
+        return out

backend/headless/fcbh_extras/chainner_models/architecture/OmniSR/OSAG.py

@@ -0,0 +1,60 @@
+#!/usr/bin/env python3
+# -*- coding:utf-8 -*-
+#############################################################
+# File: OSAG.py
+# Created Date: Tuesday April 28th 2022
+# Author: Chen Xuanhong
+# Email: chenxuanhongzju@outlook.com
+# Last Modified:  Sunday, 23rd April 2023 3:08:49 pm
+# Modified By: Chen Xuanhong
+# Copyright (c) 2020 Shanghai Jiao Tong University
+#############################################################
+
+
+import torch.nn as nn
+
+from .esa import ESA
+from .OSA import OSA_Block
+
+
+class OSAG(nn.Module):
+    def __init__(
+        self,
+        channel_num=64,
+        bias=True,
+        block_num=4,
+        ffn_bias=False,
+        window_size=0,
+        pe=False,
+    ):
+        super(OSAG, self).__init__()
+
+        # print("window_size: %d" % (window_size))
+        # print("with_pe", pe)
+        # print("ffn_bias: %d" % (ffn_bias))
+
+        # block_script_name = kwargs.get("block_script_name", "OSA")
+        # block_class_name = kwargs.get("block_class_name", "OSA_Block")
+
+        # script_name = "." + block_script_name
+        # package = __import__(script_name, fromlist=True)
+        block_class = OSA_Block  # getattr(package, block_class_name)
+        group_list = []
+        for _ in range(block_num):
+            temp_res = block_class(
+                channel_num,
+                bias,
+                ffn_bias=ffn_bias,
+                window_size=window_size,
+                with_pe=pe,
+            )
+            group_list.append(temp_res)
+        group_list.append(nn.Conv2d(channel_num, channel_num, 1, 1, 0, bias=bias))
+        self.residual_layer = nn.Sequential(*group_list)
+        esa_channel = max(channel_num // 4, 16)
+        self.esa = ESA(esa_channel, channel_num)
+
+    def forward(self, x):
+        out = self.residual_layer(x)
+        out = out + x
+        return self.esa(out)

backend/headless/fcbh_extras/chainner_models/architecture/OmniSR/OmniSR.py

@@ -0,0 +1,143 @@
+#!/usr/bin/env python3
+# -*- coding:utf-8 -*-
+#############################################################
+# File: OmniSR.py
+# Created Date: Tuesday April 28th 2022
+# Author: Chen Xuanhong
+# Email: chenxuanhongzju@outlook.com
+# Last Modified:  Sunday, 23rd April 2023 3:06:36 pm
+# Modified By: Chen Xuanhong
+# Copyright (c) 2020 Shanghai Jiao Tong University
+#############################################################
+
+import math
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from .OSAG import OSAG
+from .pixelshuffle import pixelshuffle_block
+
+
+class OmniSR(nn.Module):
+    def __init__(
+        self,
+        state_dict,
+        **kwargs,
+    ):
+        super(OmniSR, self).__init__()
+        self.state = state_dict
+
+        bias = True  # Fine to assume this for now
+        block_num = 1  # Fine to assume this for now
+        ffn_bias = True
+        pe = True
+
+        num_feat = state_dict["input.weight"].shape[0] or 64
+        num_in_ch = state_dict["input.weight"].shape[1] or 3
+        num_out_ch = num_in_ch  # we can just assume this for now. pixelshuffle smh
+
+        pixelshuffle_shape = state_dict["up.0.weight"].shape[0]
+        up_scale = math.sqrt(pixelshuffle_shape / num_out_ch)
+        if up_scale - int(up_scale) > 0:
+            print(
+                "out_nc is probably different than in_nc, scale calculation might be wrong"
+            )
+        up_scale = int(up_scale)
+        res_num = 0
+        for key in state_dict.keys():
+            if "residual_layer" in key:
+                temp_res_num = int(key.split(".")[1])
+                if temp_res_num > res_num:
+                    res_num = temp_res_num
+        res_num = res_num + 1  # zero-indexed
+
+        residual_layer = []
+        self.res_num = res_num
+
+        if (
+            "residual_layer.0.residual_layer.0.layer.2.fn.rel_pos_bias.weight"
+            in state_dict.keys()
+        ):
+            rel_pos_bias_weight = state_dict[
+                "residual_layer.0.residual_layer.0.layer.2.fn.rel_pos_bias.weight"
+            ].shape[0]
+            self.window_size = int((math.sqrt(rel_pos_bias_weight) + 1) / 2)
+        else:
+            self.window_size = 8
+
+        self.up_scale = up_scale
+
+        for _ in range(res_num):
+            temp_res = OSAG(
+                channel_num=num_feat,
+                bias=bias,
+                block_num=block_num,
+                ffn_bias=ffn_bias,
+                window_size=self.window_size,
+                pe=pe,
+            )
+            residual_layer.append(temp_res)
+        self.residual_layer = nn.Sequential(*residual_layer)
+        self.input = nn.Conv2d(
+            in_channels=num_in_ch,
+            out_channels=num_feat,
+            kernel_size=3,
+            stride=1,
+            padding=1,
+            bias=bias,
+        )
+        self.output = nn.Conv2d(
+            in_channels=num_feat,
+            out_channels=num_feat,
+            kernel_size=3,
+            stride=1,
+            padding=1,
+            bias=bias,
+        )
+        self.up = pixelshuffle_block(num_feat, num_out_ch, up_scale, bias=bias)
+
+        # self.tail   = pixelshuffle_block(num_feat,num_out_ch,up_scale,bias=bias)
+
+        # for m in self.modules():
+        #     if isinstance(m, nn.Conv2d):
+        #         n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
+        #         m.weight.data.normal_(0, sqrt(2. / n))
+
+        # chaiNNer specific stuff
+        self.model_arch = "OmniSR"
+        self.sub_type = "SR"
+        self.in_nc = num_in_ch
+        self.out_nc = num_out_ch
+        self.num_feat = num_feat
+        self.scale = up_scale
+
+        self.supports_fp16 = True  # TODO: Test this
+        self.supports_bfp16 = True
+        self.min_size_restriction = 16
+
+        self.load_state_dict(state_dict, strict=False)
+
+    def check_image_size(self, x):
+        _, _, h, w = x.size()
+        # import pdb; pdb.set_trace()
+        mod_pad_h = (self.window_size - h % self.window_size) % self.window_size
+        mod_pad_w = (self.window_size - w % self.window_size) % self.window_size
+        # x = F.pad(x, (0, mod_pad_w, 0, mod_pad_h), 'reflect')
+        x = F.pad(x, (0, mod_pad_w, 0, mod_pad_h), "constant", 0)
+        return x
+
+    def forward(self, x):
+        H, W = x.shape[2:]
+        x = self.check_image_size(x)
+
+        residual = self.input(x)
+        out = self.residual_layer(residual)
+
+        # origin
+        out = torch.add(self.output(out), residual)
+        out = self.up(out)
+
+        out = out[:, :, : H * self.up_scale, : W * self.up_scale]
+        return out

backend/headless/fcbh_extras/chainner_models/architecture/OmniSR/esa.py

@@ -0,0 +1,294 @@
+#!/usr/bin/env python3
+# -*- coding:utf-8 -*-
+#############################################################
+# File: esa.py
+# Created Date: Tuesday April 28th 2022
+# Author: Chen Xuanhong
+# Email: chenxuanhongzju@outlook.com
+# Last Modified:  Thursday, 20th April 2023 9:28:06 am
+# Modified By: Chen Xuanhong
+# Copyright (c) 2020 Shanghai Jiao Tong University
+#############################################################
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from .layernorm import LayerNorm2d
+
+
+def moment(x, dim=(2, 3), k=2):
+    assert len(x.size()) == 4
+    mean = torch.mean(x, dim=dim).unsqueeze(-1).unsqueeze(-1)
+    mk = (1 / (x.size(2) * x.size(3))) * torch.sum(torch.pow(x - mean, k), dim=dim)
+    return mk
+
+
+class ESA(nn.Module):
+    """
+    Modification of Enhanced Spatial Attention (ESA), which is proposed by
+    `Residual Feature Aggregation Network for Image Super-Resolution`
+    Note: `conv_max` and `conv3_` are NOT used here, so the corresponding codes
+    are deleted.
+    """
+
+    def __init__(self, esa_channels, n_feats, conv=nn.Conv2d):
+        super(ESA, self).__init__()
+        f = esa_channels
+        self.conv1 = conv(n_feats, f, kernel_size=1)
+        self.conv_f = conv(f, f, kernel_size=1)
+        self.conv2 = conv(f, f, kernel_size=3, stride=2, padding=0)
+        self.conv3 = conv(f, f, kernel_size=3, padding=1)
+        self.conv4 = conv(f, n_feats, kernel_size=1)
+        self.sigmoid = nn.Sigmoid()
+        self.relu = nn.ReLU(inplace=True)
+
+    def forward(self, x):
+        c1_ = self.conv1(x)
+        c1 = self.conv2(c1_)
+        v_max = F.max_pool2d(c1, kernel_size=7, stride=3)
+        c3 = self.conv3(v_max)
+        c3 = F.interpolate(
+            c3, (x.size(2), x.size(3)), mode="bilinear", align_corners=False
+        )
+        cf = self.conv_f(c1_)
+        c4 = self.conv4(c3 + cf)
+        m = self.sigmoid(c4)
+        return x * m
+
+
+class LK_ESA(nn.Module):
+    def __init__(
+        self, esa_channels, n_feats, conv=nn.Conv2d, kernel_expand=1, bias=True
+    ):
+        super(LK_ESA, self).__init__()
+        f = esa_channels
+        self.conv1 = conv(n_feats, f, kernel_size=1)
+        self.conv_f = conv(f, f, kernel_size=1)
+
+        kernel_size = 17
+        kernel_expand = kernel_expand
+        padding = kernel_size // 2
+
+        self.vec_conv = nn.Conv2d(
+            in_channels=f * kernel_expand,
+            out_channels=f * kernel_expand,
+            kernel_size=(1, kernel_size),
+            padding=(0, padding),
+            groups=2,
+            bias=bias,
+        )
+        self.vec_conv3x1 = nn.Conv2d(
+            in_channels=f * kernel_expand,
+            out_channels=f * kernel_expand,
+            kernel_size=(1, 3),
+            padding=(0, 1),
+            groups=2,
+            bias=bias,
+        )
+
+        self.hor_conv = nn.Conv2d(
+            in_channels=f * kernel_expand,
+            out_channels=f * kernel_expand,
+            kernel_size=(kernel_size, 1),
+            padding=(padding, 0),
+            groups=2,
+            bias=bias,
+        )
+        self.hor_conv1x3 = nn.Conv2d(
+            in_channels=f * kernel_expand,
+            out_channels=f * kernel_expand,
+            kernel_size=(3, 1),
+            padding=(1, 0),
+            groups=2,
+            bias=bias,
+        )
+
+        self.conv4 = conv(f, n_feats, kernel_size=1)
+        self.sigmoid = nn.Sigmoid()
+        self.relu = nn.ReLU(inplace=True)
+
+    def forward(self, x):
+        c1_ = self.conv1(x)
+
+        res = self.vec_conv(c1_) + self.vec_conv3x1(c1_)
+        res = self.hor_conv(res) + self.hor_conv1x3(res)
+
+        cf = self.conv_f(c1_)
+        c4 = self.conv4(res + cf)
+        m = self.sigmoid(c4)
+        return x * m
+
+
+class LK_ESA_LN(nn.Module):
+    def __init__(
+        self, esa_channels, n_feats, conv=nn.Conv2d, kernel_expand=1, bias=True
+    ):
+        super(LK_ESA_LN, self).__init__()
+        f = esa_channels
+        self.conv1 = conv(n_feats, f, kernel_size=1)
+        self.conv_f = conv(f, f, kernel_size=1)
+
+        kernel_size = 17
+        kernel_expand = kernel_expand
+        padding = kernel_size // 2
+
+        self.norm = LayerNorm2d(n_feats)
+
+        self.vec_conv = nn.Conv2d(
+            in_channels=f * kernel_expand,
+            out_channels=f * kernel_expand,
+            kernel_size=(1, kernel_size),
+            padding=(0, padding),
+            groups=2,
+            bias=bias,
+        )
+        self.vec_conv3x1 = nn.Conv2d(
+            in_channels=f * kernel_expand,
+            out_channels=f * kernel_expand,
+            kernel_size=(1, 3),
+            padding=(0, 1),
+            groups=2,
+            bias=bias,
+        )
+
+        self.hor_conv = nn.Conv2d(
+            in_channels=f * kernel_expand,
+            out_channels=f * kernel_expand,
+            kernel_size=(kernel_size, 1),
+            padding=(padding, 0),
+            groups=2,
+            bias=bias,
+        )
+        self.hor_conv1x3 = nn.Conv2d(
+            in_channels=f * kernel_expand,
+            out_channels=f * kernel_expand,
+            kernel_size=(3, 1),
+            padding=(1, 0),
+            groups=2,
+            bias=bias,
+        )
+
+        self.conv4 = conv(f, n_feats, kernel_size=1)
+        self.sigmoid = nn.Sigmoid()
+        self.relu = nn.ReLU(inplace=True)
+
+    def forward(self, x):
+        c1_ = self.norm(x)
+        c1_ = self.conv1(c1_)
+
+        res = self.vec_conv(c1_) + self.vec_conv3x1(c1_)
+        res = self.hor_conv(res) + self.hor_conv1x3(res)
+
+        cf = self.conv_f(c1_)
+        c4 = self.conv4(res + cf)
+        m = self.sigmoid(c4)
+        return x * m
+
+
+class AdaGuidedFilter(nn.Module):
+    def __init__(
+        self, esa_channels, n_feats, conv=nn.Conv2d, kernel_expand=1, bias=True
+    ):
+        super(AdaGuidedFilter, self).__init__()
+
+        self.gap = nn.AdaptiveAvgPool2d(1)
+        self.fc = nn.Conv2d(
+            in_channels=n_feats,
+            out_channels=1,
+            kernel_size=1,
+            padding=0,
+            stride=1,
+            groups=1,
+            bias=True,
+        )
+
+        self.r = 5
+
+    def box_filter(self, x, r):
+        channel = x.shape[1]
+        kernel_size = 2 * r + 1
+        weight = 1.0 / (kernel_size**2)
+        box_kernel = weight * torch.ones(
+            (channel, 1, kernel_size, kernel_size), dtype=torch.float32, device=x.device
+        )
+        output = F.conv2d(x, weight=box_kernel, stride=1, padding=r, groups=channel)
+        return output
+
+    def forward(self, x):
+        _, _, H, W = x.shape
+        N = self.box_filter(
+            torch.ones((1, 1, H, W), dtype=x.dtype, device=x.device), self.r
+        )
+
+        # epsilon = self.fc(self.gap(x))
+        # epsilon = torch.pow(epsilon, 2)
+        epsilon = 1e-2
+
+        mean_x = self.box_filter(x, self.r) / N
+        var_x = self.box_filter(x * x, self.r) / N - mean_x * mean_x
+
+        A = var_x / (var_x + epsilon)
+        b = (1 - A) * mean_x
+        m = A * x + b
+
+        # mean_A = self.box_filter(A, self.r) / N
+        # mean_b = self.box_filter(b, self.r) / N
+        # m = mean_A * x + mean_b
+        return x * m
+
+
+class AdaConvGuidedFilter(nn.Module):
+    def __init__(
+        self, esa_channels, n_feats, conv=nn.Conv2d, kernel_expand=1, bias=True
+    ):
+        super(AdaConvGuidedFilter, self).__init__()
+        f = esa_channels
+
+        self.conv_f = conv(f, f, kernel_size=1)
+
+        kernel_size = 17
+        kernel_expand = kernel_expand
+        padding = kernel_size // 2
+
+        self.vec_conv = nn.Conv2d(
+            in_channels=f,
+            out_channels=f,
+            kernel_size=(1, kernel_size),
+            padding=(0, padding),
+            groups=f,
+            bias=bias,
+        )
+
+        self.hor_conv = nn.Conv2d(
+            in_channels=f,
+            out_channels=f,
+            kernel_size=(kernel_size, 1),
+            padding=(padding, 0),
+            groups=f,
+            bias=bias,
+        )
+
+        self.gap = nn.AdaptiveAvgPool2d(1)
+        self.fc = nn.Conv2d(
+            in_channels=f,
+            out_channels=f,
+            kernel_size=1,
+            padding=0,
+            stride=1,
+            groups=1,
+            bias=True,
+        )
+
+    def forward(self, x):
+        y = self.vec_conv(x)
+        y = self.hor_conv(y)
+
+        sigma = torch.pow(y, 2)
+        epsilon = self.fc(self.gap(y))
+
+        weight = sigma / (sigma + epsilon)
+
+        m = weight * x + (1 - weight)
+
+        return x * m

backend/headless/fcbh_extras/chainner_models/architecture/OmniSR/layernorm.py

@@ -0,0 +1,70 @@
+#!/usr/bin/env python3
+# -*- coding:utf-8 -*-
+#############################################################
+# File: layernorm.py
+# Created Date: Tuesday April 28th 2022
+# Author: Chen Xuanhong
+# Email: chenxuanhongzju@outlook.com
+# Last Modified:  Thursday, 20th April 2023 9:28:20 am
+# Modified By: Chen Xuanhong
+# Copyright (c) 2020 Shanghai Jiao Tong University
+#############################################################
+
+import torch
+import torch.nn as nn
+
+
+class LayerNormFunction(torch.autograd.Function):
+    @staticmethod
+    def forward(ctx, x, weight, bias, eps):
+        ctx.eps = eps
+        N, C, H, W = x.size()
+        mu = x.mean(1, keepdim=True)
+        var = (x - mu).pow(2).mean(1, keepdim=True)
+        y = (x - mu) / (var + eps).sqrt()
+        ctx.save_for_backward(y, var, weight)
+        y = weight.view(1, C, 1, 1) * y + bias.view(1, C, 1, 1)
+        return y
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        eps = ctx.eps
+
+        N, C, H, W = grad_output.size()
+        y, var, weight = ctx.saved_variables
+        g = grad_output * weight.view(1, C, 1, 1)
+        mean_g = g.mean(dim=1, keepdim=True)
+
+        mean_gy = (g * y).mean(dim=1, keepdim=True)
+        gx = 1.0 / torch.sqrt(var + eps) * (g - y * mean_gy - mean_g)
+        return (
+            gx,
+            (grad_output * y).sum(dim=3).sum(dim=2).sum(dim=0),
+            grad_output.sum(dim=3).sum(dim=2).sum(dim=0),
+            None,
+        )
+
+
+class LayerNorm2d(nn.Module):
+    def __init__(self, channels, eps=1e-6):
+        super(LayerNorm2d, self).__init__()
+        self.register_parameter("weight", nn.Parameter(torch.ones(channels)))
+        self.register_parameter("bias", nn.Parameter(torch.zeros(channels)))
+        self.eps = eps
+
+    def forward(self, x):
+        return LayerNormFunction.apply(x, self.weight, self.bias, self.eps)
+
+
+class GRN(nn.Module):
+    """GRN (Global Response Normalization) layer"""
+
+    def __init__(self, dim):
+        super().__init__()
+        self.gamma = nn.Parameter(torch.zeros(1, dim, 1, 1))
+        self.beta = nn.Parameter(torch.zeros(1, dim, 1, 1))
+
+    def forward(self, x):
+        Gx = torch.norm(x, p=2, dim=(2, 3), keepdim=True)
+        Nx = Gx / (Gx.mean(dim=1, keepdim=True) + 1e-6)
+        return self.gamma * (x * Nx) + self.beta + x

backend/headless/fcbh_extras/chainner_models/architecture/OmniSR/pixelshuffle.py

@@ -0,0 +1,31 @@
+#!/usr/bin/env python3
+# -*- coding:utf-8 -*-
+#############################################################
+# File: pixelshuffle.py
+# Created Date: Friday July 1st 2022
+# Author: Chen Xuanhong
+# Email: chenxuanhongzju@outlook.com
+# Last Modified:  Friday, 1st July 2022 10:18:39 am
+# Modified By: Chen Xuanhong
+# Copyright (c) 2022 Shanghai Jiao Tong University
+#############################################################
+
+import torch.nn as nn
+
+
+def pixelshuffle_block(
+    in_channels, out_channels, upscale_factor=2, kernel_size=3, bias=False
+):
+    """
+    Upsample features according to `upscale_factor`.
+    """
+    padding = kernel_size // 2
+    conv = nn.Conv2d(
+        in_channels,
+        out_channels * (upscale_factor**2),
+        kernel_size,
+        padding=1,
+        bias=bias,
+    )
+    pixel_shuffle = nn.PixelShuffle(upscale_factor)
+    return nn.Sequential(*[conv, pixel_shuffle])

backend/headless/fcbh_extras/chainner_models/architecture/RRDB.py

@@ -0,0 +1,296 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+import functools
+import math
+import re
+from collections import OrderedDict
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from . import block as B
+
+
+# Borrowed from https://github.com/rlaphoenix/VSGAN/blob/master/vsgan/archs/ESRGAN.py
+# Which enhanced stuff that was already here
+class RRDBNet(nn.Module):
+    def __init__(
+        self,
+        state_dict,
+        norm=None,
+        act: str = "leakyrelu",
+        upsampler: str = "upconv",
+        mode: B.ConvMode = "CNA",
+    ) -> None:
+        """
+        ESRGAN - Enhanced Super-Resolution Generative Adversarial Networks.
+        By Xintao Wang, Ke Yu, Shixiang Wu, Jinjin Gu, Yihao Liu, Chao Dong, Yu Qiao,
+        and Chen Change Loy.
+        This is old-arch Residual in Residual Dense Block Network and is not
+        the newest revision that's available at github.com/xinntao/ESRGAN.
+        This is on purpose, the newest Network has severely limited the
+        potential use of the Network with no benefits.
+        This network supports model files from both new and old-arch.
+        Args:
+            norm: Normalization layer
+            act: Activation layer
+            upsampler: Upsample layer. upconv, pixel_shuffle
+            mode: Convolution mode
+        """
+        super(RRDBNet, self).__init__()
+        self.model_arch = "ESRGAN"
+        self.sub_type = "SR"
+
+        self.state = state_dict
+        self.norm = norm
+        self.act = act
+        self.upsampler = upsampler
+        self.mode = mode
+
+        self.state_map = {
+            # currently supports old, new, and newer RRDBNet arch models
+            # ESRGAN, BSRGAN/RealSR, Real-ESRGAN
+            "model.0.weight": ("conv_first.weight",),
+            "model.0.bias": ("conv_first.bias",),
+            "model.1.sub./NB/.weight": ("trunk_conv.weight", "conv_body.weight"),
+            "model.1.sub./NB/.bias": ("trunk_conv.bias", "conv_body.bias"),
+            r"model.1.sub.\1.RDB\2.conv\3.0.\4": (
+                r"RRDB_trunk\.(\d+)\.RDB(\d)\.conv(\d+)\.(weight|bias)",
+                r"body\.(\d+)\.rdb(\d)\.conv(\d+)\.(weight|bias)",
+            ),
+        }
+        if "params_ema" in self.state:
+            self.state = self.state["params_ema"]
+            # self.model_arch = "RealESRGAN"
+        self.num_blocks = self.get_num_blocks()
+        self.plus = any("conv1x1" in k for k in self.state.keys())
+        if self.plus:
+            self.model_arch = "ESRGAN+"
+
+        self.state = self.new_to_old_arch(self.state)
+
+        self.key_arr = list(self.state.keys())
+
+        self.in_nc: int = self.state[self.key_arr[0]].shape[1]
+        self.out_nc: int = self.state[self.key_arr[-1]].shape[0]
+
+        self.scale: int = self.get_scale()
+        self.num_filters: int = self.state[self.key_arr[0]].shape[0]
+
+        c2x2 = False
+        if self.state["model.0.weight"].shape[-2] == 2:
+            c2x2 = True
+            self.scale = round(math.sqrt(self.scale / 4))
+            self.model_arch = "ESRGAN-2c2"
+
+        self.supports_fp16 = True
+        self.supports_bfp16 = True
+        self.min_size_restriction = None
+
+        # Detect if pixelunshuffle was used (Real-ESRGAN)
+        if self.in_nc in (self.out_nc * 4, self.out_nc * 16) and self.out_nc in (
+            self.in_nc / 4,
+            self.in_nc / 16,
+        ):
+            self.shuffle_factor = int(math.sqrt(self.in_nc / self.out_nc))
+        else:
+            self.shuffle_factor = None
+
+        upsample_block = {
+            "upconv": B.upconv_block,
+            "pixel_shuffle": B.pixelshuffle_block,
+        }.get(self.upsampler)
+        if upsample_block is None:
+            raise NotImplementedError(f"Upsample mode [{self.upsampler}] is not found")
+
+        if self.scale == 3:
+            upsample_blocks = upsample_block(
+                in_nc=self.num_filters,
+                out_nc=self.num_filters,
+                upscale_factor=3,
+                act_type=self.act,
+                c2x2=c2x2,
+            )
+        else:
+            upsample_blocks = [
+                upsample_block(
+                    in_nc=self.num_filters,
+                    out_nc=self.num_filters,
+                    act_type=self.act,
+                    c2x2=c2x2,
+                )
+                for _ in range(int(math.log(self.scale, 2)))
+            ]
+
+        self.model = B.sequential(
+            # fea conv
+            B.conv_block(
+                in_nc=self.in_nc,
+                out_nc=self.num_filters,
+                kernel_size=3,
+                norm_type=None,
+                act_type=None,
+                c2x2=c2x2,
+            ),
+            B.ShortcutBlock(
+                B.sequential(
+                    # rrdb blocks
+                    *[
+                        B.RRDB(
+                            nf=self.num_filters,
+                            kernel_size=3,
+                            gc=32,
+                            stride=1,
+                            bias=True,
+                            pad_type="zero",
+                            norm_type=self.norm,
+                            act_type=self.act,
+                            mode="CNA",
+                            plus=self.plus,
+                            c2x2=c2x2,
+                        )
+                        for _ in range(self.num_blocks)
+                    ],
+                    # lr conv
+                    B.conv_block(
+                        in_nc=self.num_filters,
+                        out_nc=self.num_filters,
+                        kernel_size=3,
+                        norm_type=self.norm,
+                        act_type=None,
+                        mode=self.mode,
+                        c2x2=c2x2,
+                    ),
+                )
+            ),
+            *upsample_blocks,
+            # hr_conv0
+            B.conv_block(
+                in_nc=self.num_filters,
+                out_nc=self.num_filters,
+                kernel_size=3,
+                norm_type=None,
+                act_type=self.act,
+                c2x2=c2x2,
+            ),
+            # hr_conv1
+            B.conv_block(
+                in_nc=self.num_filters,
+                out_nc=self.out_nc,
+                kernel_size=3,
+                norm_type=None,
+                act_type=None,
+                c2x2=c2x2,
+            ),
+        )
+
+        # Adjust these properties for calculations outside of the model
+        if self.shuffle_factor:
+            self.in_nc //= self.shuffle_factor**2
+            self.scale //= self.shuffle_factor
+
+        self.load_state_dict(self.state, strict=False)
+
+    def new_to_old_arch(self, state):
+        """Convert a new-arch model state dictionary to an old-arch dictionary."""
+        if "params_ema" in state:
+            state = state["params_ema"]
+
+        if "conv_first.weight" not in state:
+            # model is already old arch, this is a loose check, but should be sufficient
+            return state
+
+        # add nb to state keys
+        for kind in ("weight", "bias"):
+            self.state_map[f"model.1.sub.{self.num_blocks}.{kind}"] = self.state_map[
+                f"model.1.sub./NB/.{kind}"
+            ]
+            del self.state_map[f"model.1.sub./NB/.{kind}"]
+
+        old_state = OrderedDict()
+        for old_key, new_keys in self.state_map.items():
+            for new_key in new_keys:
+                if r"\1" in old_key:
+                    for k, v in state.items():
+                        sub = re.sub(new_key, old_key, k)
+                        if sub != k:
+                            old_state[sub] = v
+                else:
+                    if new_key in state:
+                        old_state[old_key] = state[new_key]
+
+        # upconv layers
+        max_upconv = 0
+        for key in state.keys():
+            match = re.match(r"(upconv|conv_up)(\d)\.(weight|bias)", key)
+            if match is not None:
+                _, key_num, key_type = match.groups()
+                old_state[f"model.{int(key_num) * 3}.{key_type}"] = state[key]
+                max_upconv = max(max_upconv, int(key_num) * 3)
+
+        # final layers
+        for key in state.keys():
+            if key in ("HRconv.weight", "conv_hr.weight"):
+                old_state[f"model.{max_upconv + 2}.weight"] = state[key]
+            elif key in ("HRconv.bias", "conv_hr.bias"):
+                old_state[f"model.{max_upconv + 2}.bias"] = state[key]
+            elif key in ("conv_last.weight",):
+                old_state[f"model.{max_upconv + 4}.weight"] = state[key]
+            elif key in ("conv_last.bias",):
+                old_state[f"model.{max_upconv + 4}.bias"] = state[key]
+
+        # Sort by first numeric value of each layer
+        def compare(item1, item2):
+            parts1 = item1.split(".")
+            parts2 = item2.split(".")
+            int1 = int(parts1[1])
+            int2 = int(parts2[1])
+            return int1 - int2
+
+        sorted_keys = sorted(old_state.keys(), key=functools.cmp_to_key(compare))
+
+        # Rebuild the output dict in the right order
+        out_dict = OrderedDict((k, old_state[k]) for k in sorted_keys)
+
+        return out_dict
+
+    def get_scale(self, min_part: int = 6) -> int:
+        n = 0
+        for part in list(self.state):
+            parts = part.split(".")[1:]
+            if len(parts) == 2:
+                part_num = int(parts[0])
+                if part_num > min_part and parts[1] == "weight":
+                    n += 1
+        return 2**n
+
+    def get_num_blocks(self) -> int:
+        nbs = []
+        state_keys = self.state_map[r"model.1.sub.\1.RDB\2.conv\3.0.\4"] + (
+            r"model\.\d+\.sub\.(\d+)\.RDB(\d+)\.conv(\d+)\.0\.(weight|bias)",
+        )
+        for state_key in state_keys:
+            for k in self.state:
+                m = re.search(state_key, k)
+                if m:
+                    nbs.append(int(m.group(1)))
+            if nbs:
+                break
+        return max(*nbs) + 1
+
+    def forward(self, x):
+        if self.shuffle_factor:
+            _, _, h, w = x.size()
+            mod_pad_h = (
+                self.shuffle_factor - h % self.shuffle_factor
+            ) % self.shuffle_factor
+            mod_pad_w = (
+                self.shuffle_factor - w % self.shuffle_factor
+            ) % self.shuffle_factor
+            x = F.pad(x, (0, mod_pad_w, 0, mod_pad_h), "reflect")
+            x = torch.pixel_unshuffle(x, downscale_factor=self.shuffle_factor)
+            x = self.model(x)
+            return x[:, :, : h * self.scale, : w * self.scale]
+        return self.model(x)

backend/headless/fcbh_extras/chainner_models/architecture/SCUNet.py

@@ -0,0 +1,455 @@
+# pylint: skip-file
+# -----------------------------------------------------------------------------------
+# SCUNet: Practical Blind Denoising via Swin-Conv-UNet and Data Synthesis, https://arxiv.org/abs/2203.13278
+# Zhang, Kai and Li, Yawei and Liang, Jingyun and Cao, Jiezhang and Zhang, Yulun and Tang, Hao and Timofte, Radu and Van Gool, Luc
+# -----------------------------------------------------------------------------------
+
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from einops import rearrange
+from einops.layers.torch import Rearrange
+
+from .timm.drop import DropPath
+from .timm.weight_init import trunc_normal_
+
+
+# Borrowed from https://github.com/cszn/SCUNet/blob/main/models/network_scunet.py
+class WMSA(nn.Module):
+    """Self-attention module in Swin Transformer"""
+
+    def __init__(self, input_dim, output_dim, head_dim, window_size, type):
+        super(WMSA, self).__init__()
+        self.input_dim = input_dim
+        self.output_dim = output_dim
+        self.head_dim = head_dim
+        self.scale = self.head_dim**-0.5
+        self.n_heads = input_dim // head_dim
+        self.window_size = window_size
+        self.type = type
+        self.embedding_layer = nn.Linear(self.input_dim, 3 * self.input_dim, bias=True)
+
+        self.relative_position_params = nn.Parameter(
+            torch.zeros((2 * window_size - 1) * (2 * window_size - 1), self.n_heads)
+        )
+        # TODO recover
+        # self.relative_position_params = nn.Parameter(torch.zeros(self.n_heads, 2 * window_size - 1, 2 * window_size -1))
+        self.relative_position_params = nn.Parameter(
+            torch.zeros((2 * window_size - 1) * (2 * window_size - 1), self.n_heads)
+        )
+
+        self.linear = nn.Linear(self.input_dim, self.output_dim)
+
+        trunc_normal_(self.relative_position_params, std=0.02)
+        self.relative_position_params = torch.nn.Parameter(
+            self.relative_position_params.view(
+                2 * window_size - 1, 2 * window_size - 1, self.n_heads
+            )
+            .transpose(1, 2)
+            .transpose(0, 1)
+        )
+
+    def generate_mask(self, h, w, p, shift):
+        """generating the mask of SW-MSA
+        Args:
+            shift: shift parameters in CyclicShift.
+        Returns:
+            attn_mask: should be (1 1 w p p),
+        """
+        # supporting square.
+        attn_mask = torch.zeros(
+            h,
+            w,
+            p,
+            p,
+            p,
+            p,
+            dtype=torch.bool,
+            device=self.relative_position_params.device,
+        )
+        if self.type == "W":
+            return attn_mask
+
+        s = p - shift
+        attn_mask[-1, :, :s, :, s:, :] = True
+        attn_mask[-1, :, s:, :, :s, :] = True
+        attn_mask[:, -1, :, :s, :, s:] = True
+        attn_mask[:, -1, :, s:, :, :s] = True
+        attn_mask = rearrange(
+            attn_mask, "w1 w2 p1 p2 p3 p4 -> 1 1 (w1 w2) (p1 p2) (p3 p4)"
+        )
+        return attn_mask
+
+    def forward(self, x):
+        """Forward pass of Window Multi-head Self-attention module.
+        Args:
+            x: input tensor with shape of [b h w c];
+            attn_mask: attention mask, fill -inf where the value is True;
+        Returns:
+            output: tensor shape [b h w c]
+        """
+        if self.type != "W":
+            x = torch.roll(
+                x,
+                shifts=(-(self.window_size // 2), -(self.window_size // 2)),
+                dims=(1, 2),
+            )
+
+        x = rearrange(
+            x,
+            "b (w1 p1) (w2 p2) c -> b w1 w2 p1 p2 c",
+            p1=self.window_size,
+            p2=self.window_size,
+        )
+        h_windows = x.size(1)
+        w_windows = x.size(2)
+        # square validation
+        # assert h_windows == w_windows
+
+        x = rearrange(
+            x,
+            "b w1 w2 p1 p2 c -> b (w1 w2) (p1 p2) c",
+            p1=self.window_size,
+            p2=self.window_size,
+        )
+        qkv = self.embedding_layer(x)
+        q, k, v = rearrange(
+            qkv, "b nw np (threeh c) -> threeh b nw np c", c=self.head_dim
+        ).chunk(3, dim=0)
+        sim = torch.einsum("hbwpc,hbwqc->hbwpq", q, k) * self.scale
+        # Adding learnable relative embedding
+        sim = sim + rearrange(self.relative_embedding(), "h p q -> h 1 1 p q")
+        # Using Attn Mask to distinguish different subwindows.
+        if self.type != "W":
+            attn_mask = self.generate_mask(
+                h_windows, w_windows, self.window_size, shift=self.window_size // 2
+            )
+            sim = sim.masked_fill_(attn_mask, float("-inf"))
+
+        probs = nn.functional.softmax(sim, dim=-1)
+        output = torch.einsum("hbwij,hbwjc->hbwic", probs, v)
+        output = rearrange(output, "h b w p c -> b w p (h c)")
+        output = self.linear(output)
+        output = rearrange(
+            output,
+            "b (w1 w2) (p1 p2) c -> b (w1 p1) (w2 p2) c",
+            w1=h_windows,
+            p1=self.window_size,
+        )
+
+        if self.type != "W":
+            output = torch.roll(
+                output,
+                shifts=(self.window_size // 2, self.window_size // 2),
+                dims=(1, 2),
+            )
+
+        return output
+
+    def relative_embedding(self):
+        cord = torch.tensor(
+            np.array(
+                [
+                    [i, j]
+                    for i in range(self.window_size)
+                    for j in range(self.window_size)
+                ]
+            )
+        )
+        relation = cord[:, None, :] - cord[None, :, :] + self.window_size - 1
+        # negative is allowed
+        return self.relative_position_params[
+            :, relation[:, :, 0].long(), relation[:, :, 1].long()
+        ]
+
+
+class Block(nn.Module):
+    def __init__(
+        self,
+        input_dim,
+        output_dim,
+        head_dim,
+        window_size,
+        drop_path,
+        type="W",
+        input_resolution=None,
+    ):
+        """SwinTransformer Block"""
+        super(Block, self).__init__()
+        self.input_dim = input_dim
+        self.output_dim = output_dim
+        assert type in ["W", "SW"]
+        self.type = type
+        if input_resolution <= window_size:
+            self.type = "W"
+
+        self.ln1 = nn.LayerNorm(input_dim)
+        self.msa = WMSA(input_dim, input_dim, head_dim, window_size, self.type)
+        self.drop_path = DropPath(drop_path) if drop_path > 0.0 else nn.Identity()
+        self.ln2 = nn.LayerNorm(input_dim)
+        self.mlp = nn.Sequential(
+            nn.Linear(input_dim, 4 * input_dim),
+            nn.GELU(),
+            nn.Linear(4 * input_dim, output_dim),
+        )
+
+    def forward(self, x):
+        x = x + self.drop_path(self.msa(self.ln1(x)))
+        x = x + self.drop_path(self.mlp(self.ln2(x)))
+        return x
+
+
+class ConvTransBlock(nn.Module):
+    def __init__(
+        self,
+        conv_dim,
+        trans_dim,
+        head_dim,
+        window_size,
+        drop_path,
+        type="W",
+        input_resolution=None,
+    ):
+        """SwinTransformer and Conv Block"""
+        super(ConvTransBlock, self).__init__()
+        self.conv_dim = conv_dim
+        self.trans_dim = trans_dim
+        self.head_dim = head_dim
+        self.window_size = window_size
+        self.drop_path = drop_path
+        self.type = type
+        self.input_resolution = input_resolution
+
+        assert self.type in ["W", "SW"]
+        if self.input_resolution <= self.window_size:
+            self.type = "W"
+
+        self.trans_block = Block(
+            self.trans_dim,
+            self.trans_dim,
+            self.head_dim,
+            self.window_size,
+            self.drop_path,
+            self.type,
+            self.input_resolution,
+        )
+        self.conv1_1 = nn.Conv2d(
+            self.conv_dim + self.trans_dim,
+            self.conv_dim + self.trans_dim,
+            1,
+            1,
+            0,
+            bias=True,
+        )
+        self.conv1_2 = nn.Conv2d(
+            self.conv_dim + self.trans_dim,
+            self.conv_dim + self.trans_dim,
+            1,
+            1,
+            0,
+            bias=True,
+        )
+
+        self.conv_block = nn.Sequential(
+            nn.Conv2d(self.conv_dim, self.conv_dim, 3, 1, 1, bias=False),
+            nn.ReLU(True),
+            nn.Conv2d(self.conv_dim, self.conv_dim, 3, 1, 1, bias=False),
+        )
+
+    def forward(self, x):
+        conv_x, trans_x = torch.split(
+            self.conv1_1(x), (self.conv_dim, self.trans_dim), dim=1
+        )
+        conv_x = self.conv_block(conv_x) + conv_x
+        trans_x = Rearrange("b c h w -> b h w c")(trans_x)
+        trans_x = self.trans_block(trans_x)
+        trans_x = Rearrange("b h w c -> b c h w")(trans_x)
+        res = self.conv1_2(torch.cat((conv_x, trans_x), dim=1))
+        x = x + res
+
+        return x
+
+
+class SCUNet(nn.Module):
+    def __init__(
+        self,
+        state_dict,
+        in_nc=3,
+        config=[4, 4, 4, 4, 4, 4, 4],
+        dim=64,
+        drop_path_rate=0.0,
+        input_resolution=256,
+    ):
+        super(SCUNet, self).__init__()
+        self.model_arch = "SCUNet"
+        self.sub_type = "SR"
+
+        self.num_filters: int = 0
+
+        self.state = state_dict
+        self.config = config
+        self.dim = dim
+        self.head_dim = 32
+        self.window_size = 8
+
+        self.in_nc = in_nc
+        self.out_nc = self.in_nc
+        self.scale = 1
+        self.supports_fp16 = True
+
+        # drop path rate for each layer
+        dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(config))]
+
+        self.m_head = [nn.Conv2d(in_nc, dim, 3, 1, 1, bias=False)]
+
+        begin = 0
+        self.m_down1 = [
+            ConvTransBlock(
+                dim // 2,
+                dim // 2,
+                self.head_dim,
+                self.window_size,
+                dpr[i + begin],
+                "W" if not i % 2 else "SW",
+                input_resolution,
+            )
+            for i in range(config[0])
+        ] + [nn.Conv2d(dim, 2 * dim, 2, 2, 0, bias=False)]
+
+        begin += config[0]
+        self.m_down2 = [
+            ConvTransBlock(
+                dim,
+                dim,
+                self.head_dim,
+                self.window_size,
+                dpr[i + begin],
+                "W" if not i % 2 else "SW",
+                input_resolution // 2,
+            )
+            for i in range(config[1])
+        ] + [nn.Conv2d(2 * dim, 4 * dim, 2, 2, 0, bias=False)]
+
+        begin += config[1]
+        self.m_down3 = [
+            ConvTransBlock(
+                2 * dim,
+                2 * dim,
+                self.head_dim,
+                self.window_size,
+                dpr[i + begin],
+                "W" if not i % 2 else "SW",
+                input_resolution // 4,
+            )
+            for i in range(config[2])
+        ] + [nn.Conv2d(4 * dim, 8 * dim, 2, 2, 0, bias=False)]
+
+        begin += config[2]
+        self.m_body = [
+            ConvTransBlock(
+                4 * dim,
+                4 * dim,
+                self.head_dim,
+                self.window_size,
+                dpr[i + begin],
+                "W" if not i % 2 else "SW",
+                input_resolution // 8,
+            )
+            for i in range(config[3])
+        ]
+
+        begin += config[3]
+        self.m_up3 = [
+            nn.ConvTranspose2d(8 * dim, 4 * dim, 2, 2, 0, bias=False),
+        ] + [
+            ConvTransBlock(
+                2 * dim,
+                2 * dim,
+                self.head_dim,
+                self.window_size,
+                dpr[i + begin],
+                "W" if not i % 2 else "SW",
+                input_resolution // 4,
+            )
+            for i in range(config[4])
+        ]
+
+        begin += config[4]
+        self.m_up2 = [
+            nn.ConvTranspose2d(4 * dim, 2 * dim, 2, 2, 0, bias=False),
+        ] + [
+            ConvTransBlock(
+                dim,
+                dim,
+                self.head_dim,
+                self.window_size,
+                dpr[i + begin],
+                "W" if not i % 2 else "SW",
+                input_resolution // 2,
+            )
+            for i in range(config[5])
+        ]
+
+        begin += config[5]
+        self.m_up1 = [
+            nn.ConvTranspose2d(2 * dim, dim, 2, 2, 0, bias=False),
+        ] + [
+            ConvTransBlock(
+                dim // 2,
+                dim // 2,
+                self.head_dim,
+                self.window_size,
+                dpr[i + begin],
+                "W" if not i % 2 else "SW",
+                input_resolution,
+            )
+            for i in range(config[6])
+        ]
+
+        self.m_tail = [nn.Conv2d(dim, in_nc, 3, 1, 1, bias=False)]
+
+        self.m_head = nn.Sequential(*self.m_head)
+        self.m_down1 = nn.Sequential(*self.m_down1)
+        self.m_down2 = nn.Sequential(*self.m_down2)
+        self.m_down3 = nn.Sequential(*self.m_down3)
+        self.m_body = nn.Sequential(*self.m_body)
+        self.m_up3 = nn.Sequential(*self.m_up3)
+        self.m_up2 = nn.Sequential(*self.m_up2)
+        self.m_up1 = nn.Sequential(*self.m_up1)
+        self.m_tail = nn.Sequential(*self.m_tail)
+        # self.apply(self._init_weights)
+        self.load_state_dict(state_dict, strict=True)
+
+    def check_image_size(self, x):
+        _, _, h, w = x.size()
+        mod_pad_h = (64 - h % 64) % 64
+        mod_pad_w = (64 - w % 64) % 64
+        x = F.pad(x, (0, mod_pad_w, 0, mod_pad_h), "reflect")
+        return x
+
+    def forward(self, x0):
+        h, w = x0.size()[-2:]
+        x0 = self.check_image_size(x0)
+
+        x1 = self.m_head(x0)
+        x2 = self.m_down1(x1)
+        x3 = self.m_down2(x2)
+        x4 = self.m_down3(x3)
+        x = self.m_body(x4)
+        x = self.m_up3(x + x4)
+        x = self.m_up2(x + x3)
+        x = self.m_up1(x + x2)
+        x = self.m_tail(x + x1)
+
+        x = x[:, :, :h, :w]
+        return x
+
+    def _init_weights(self, m):
+        if isinstance(m, nn.Linear):
+            trunc_normal_(m.weight, std=0.02)
+            if m.bias is not None:
+                nn.init.constant_(m.bias, 0)
+        elif isinstance(m, nn.LayerNorm):
+            nn.init.constant_(m.bias, 0)
+            nn.init.constant_(m.weight, 1.0)

backend/headless/fcbh_extras/chainner_models/architecture/SPSR.py

@@ -0,0 +1,383 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+import math
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from . import block as B
+
+
+class Get_gradient_nopadding(nn.Module):
+    def __init__(self):
+        super(Get_gradient_nopadding, self).__init__()
+        kernel_v = [[0, -1, 0], [0, 0, 0], [0, 1, 0]]
+        kernel_h = [[0, 0, 0], [-1, 0, 1], [0, 0, 0]]
+        kernel_h = torch.FloatTensor(kernel_h).unsqueeze(0).unsqueeze(0)
+        kernel_v = torch.FloatTensor(kernel_v).unsqueeze(0).unsqueeze(0)
+        self.weight_h = nn.Parameter(data=kernel_h, requires_grad=False)  # type: ignore
+
+        self.weight_v = nn.Parameter(data=kernel_v, requires_grad=False)  # type: ignore
+
+    def forward(self, x):
+        x_list = []
+        for i in range(x.shape[1]):
+            x_i = x[:, i]
+            x_i_v = F.conv2d(x_i.unsqueeze(1), self.weight_v, padding=1)
+            x_i_h = F.conv2d(x_i.unsqueeze(1), self.weight_h, padding=1)
+            x_i = torch.sqrt(torch.pow(x_i_v, 2) + torch.pow(x_i_h, 2) + 1e-6)
+            x_list.append(x_i)
+
+        x = torch.cat(x_list, dim=1)
+
+        return x
+
+
+class SPSRNet(nn.Module):
+    def __init__(
+        self,
+        state_dict,
+        norm=None,
+        act: str = "leakyrelu",
+        upsampler: str = "upconv",
+        mode: B.ConvMode = "CNA",
+    ):
+        super(SPSRNet, self).__init__()
+        self.model_arch = "SPSR"
+        self.sub_type = "SR"
+
+        self.state = state_dict
+        self.norm = norm
+        self.act = act
+        self.upsampler = upsampler
+        self.mode = mode
+
+        self.num_blocks = self.get_num_blocks()
+
+        self.in_nc: int = self.state["model.0.weight"].shape[1]
+        self.out_nc: int = self.state["f_HR_conv1.0.bias"].shape[0]
+
+        self.scale = self.get_scale(4)
+        self.num_filters: int = self.state["model.0.weight"].shape[0]
+
+        self.supports_fp16 = True
+        self.supports_bfp16 = True
+        self.min_size_restriction = None
+
+        n_upscale = int(math.log(self.scale, 2))
+        if self.scale == 3:
+            n_upscale = 1
+
+        fea_conv = B.conv_block(
+            self.in_nc, self.num_filters, kernel_size=3, norm_type=None, act_type=None
+        )
+        rb_blocks = [
+            B.RRDB(
+                self.num_filters,
+                kernel_size=3,
+                gc=32,
+                stride=1,
+                bias=True,
+                pad_type="zero",
+                norm_type=norm,
+                act_type=act,
+                mode="CNA",
+            )
+            for _ in range(self.num_blocks)
+        ]
+        LR_conv = B.conv_block(
+            self.num_filters,
+            self.num_filters,
+            kernel_size=3,
+            norm_type=norm,
+            act_type=None,
+            mode=mode,
+        )
+
+        if upsampler == "upconv":
+            upsample_block = B.upconv_block
+        elif upsampler == "pixelshuffle":
+            upsample_block = B.pixelshuffle_block
+        else:
+            raise NotImplementedError(f"upsample mode [{upsampler}] is not found")
+        if self.scale == 3:
+            a_upsampler = upsample_block(
+                self.num_filters, self.num_filters, 3, act_type=act
+            )
+        else:
+            a_upsampler = [
+                upsample_block(self.num_filters, self.num_filters, act_type=act)
+                for _ in range(n_upscale)
+            ]
+        self.HR_conv0_new = B.conv_block(
+            self.num_filters,
+            self.num_filters,
+            kernel_size=3,
+            norm_type=None,
+            act_type=act,
+        )
+        self.HR_conv1_new = B.conv_block(
+            self.num_filters,
+            self.num_filters,
+            kernel_size=3,
+            norm_type=None,
+            act_type=None,
+        )
+
+        self.model = B.sequential(
+            fea_conv,
+            B.ShortcutBlockSPSR(B.sequential(*rb_blocks, LR_conv)),
+            *a_upsampler,
+            self.HR_conv0_new,
+        )
+
+        self.get_g_nopadding = Get_gradient_nopadding()
+
+        self.b_fea_conv = B.conv_block(
+            self.in_nc, self.num_filters, kernel_size=3, norm_type=None, act_type=None
+        )
+
+        self.b_concat_1 = B.conv_block(
+            2 * self.num_filters,
+            self.num_filters,
+            kernel_size=3,
+            norm_type=None,
+            act_type=None,
+        )
+        self.b_block_1 = B.RRDB(
+            self.num_filters * 2,
+            kernel_size=3,
+            gc=32,
+            stride=1,
+            bias=True,
+            pad_type="zero",
+            norm_type=norm,
+            act_type=act,
+            mode="CNA",
+        )
+
+        self.b_concat_2 = B.conv_block(
+            2 * self.num_filters,
+            self.num_filters,
+            kernel_size=3,
+            norm_type=None,
+            act_type=None,
+        )
+        self.b_block_2 = B.RRDB(
+            self.num_filters * 2,
+            kernel_size=3,
+            gc=32,
+            stride=1,
+            bias=True,
+            pad_type="zero",
+            norm_type=norm,
+            act_type=act,
+            mode="CNA",
+        )
+
+        self.b_concat_3 = B.conv_block(
+            2 * self.num_filters,
+            self.num_filters,
+            kernel_size=3,
+            norm_type=None,
+            act_type=None,
+        )
+        self.b_block_3 = B.RRDB(
+            self.num_filters * 2,
+            kernel_size=3,
+            gc=32,
+            stride=1,
+            bias=True,
+            pad_type="zero",
+            norm_type=norm,
+            act_type=act,
+            mode="CNA",
+        )
+
+        self.b_concat_4 = B.conv_block(
+            2 * self.num_filters,
+            self.num_filters,
+            kernel_size=3,
+            norm_type=None,
+            act_type=None,
+        )
+        self.b_block_4 = B.RRDB(
+            self.num_filters * 2,
+            kernel_size=3,
+            gc=32,
+            stride=1,
+            bias=True,
+            pad_type="zero",
+            norm_type=norm,
+            act_type=act,
+            mode="CNA",
+        )
+
+        self.b_LR_conv = B.conv_block(
+            self.num_filters,
+            self.num_filters,
+            kernel_size=3,
+            norm_type=norm,
+            act_type=None,
+            mode=mode,
+        )
+
+        if upsampler == "upconv":
+            upsample_block = B.upconv_block
+        elif upsampler == "pixelshuffle":
+            upsample_block = B.pixelshuffle_block
+        else:
+            raise NotImplementedError(f"upsample mode [{upsampler}] is not found")
+        if self.scale == 3:
+            b_upsampler = upsample_block(
+                self.num_filters, self.num_filters, 3, act_type=act
+            )
+        else:
+            b_upsampler = [
+                upsample_block(self.num_filters, self.num_filters, act_type=act)
+                for _ in range(n_upscale)
+            ]
+
+        b_HR_conv0 = B.conv_block(
+            self.num_filters,
+            self.num_filters,
+            kernel_size=3,
+            norm_type=None,
+            act_type=act,
+        )
+        b_HR_conv1 = B.conv_block(
+            self.num_filters,
+            self.num_filters,
+            kernel_size=3,
+            norm_type=None,
+            act_type=None,
+        )
+
+        self.b_module = B.sequential(*b_upsampler, b_HR_conv0, b_HR_conv1)
+
+        self.conv_w = B.conv_block(
+            self.num_filters, self.out_nc, kernel_size=1, norm_type=None, act_type=None
+        )
+
+        self.f_concat = B.conv_block(
+            self.num_filters * 2,
+            self.num_filters,
+            kernel_size=3,
+            norm_type=None,
+            act_type=None,
+        )
+
+        self.f_block = B.RRDB(
+            self.num_filters * 2,
+            kernel_size=3,
+            gc=32,
+            stride=1,
+            bias=True,
+            pad_type="zero",
+            norm_type=norm,
+            act_type=act,
+            mode="CNA",
+        )
+
+        self.f_HR_conv0 = B.conv_block(
+            self.num_filters,
+            self.num_filters,
+            kernel_size=3,
+            norm_type=None,
+            act_type=act,
+        )
+        self.f_HR_conv1 = B.conv_block(
+            self.num_filters, self.out_nc, kernel_size=3, norm_type=None, act_type=None
+        )
+
+        self.load_state_dict(self.state, strict=False)
+
+    def get_scale(self, min_part: int = 4) -> int:
+        n = 0
+        for part in list(self.state):
+            parts = part.split(".")
+            if len(parts) == 3:
+                part_num = int(parts[1])
+                if part_num > min_part and parts[0] == "model" and parts[2] == "weight":
+                    n += 1
+        return 2**n
+
+    def get_num_blocks(self) -> int:
+        nb = 0
+        for part in list(self.state):
+            parts = part.split(".")
+            n_parts = len(parts)
+            if n_parts == 5 and parts[2] == "sub":
+                nb = int(parts[3])
+        return nb
+
+    def forward(self, x):
+        x_grad = self.get_g_nopadding(x)
+        x = self.model[0](x)
+
+        x, block_list = self.model[1](x)
+
+        x_ori = x
+        for i in range(5):
+            x = block_list[i](x)
+        x_fea1 = x
+
+        for i in range(5):
+            x = block_list[i + 5](x)
+        x_fea2 = x
+
+        for i in range(5):
+            x = block_list[i + 10](x)
+        x_fea3 = x
+
+        for i in range(5):
+            x = block_list[i + 15](x)
+        x_fea4 = x
+
+        x = block_list[20:](x)
+        # short cut
+        x = x_ori + x
+        x = self.model[2:](x)
+        x = self.HR_conv1_new(x)
+
+        x_b_fea = self.b_fea_conv(x_grad)
+        x_cat_1 = torch.cat([x_b_fea, x_fea1], dim=1)
+
+        x_cat_1 = self.b_block_1(x_cat_1)
+        x_cat_1 = self.b_concat_1(x_cat_1)
+
+        x_cat_2 = torch.cat([x_cat_1, x_fea2], dim=1)
+
+        x_cat_2 = self.b_block_2(x_cat_2)
+        x_cat_2 = self.b_concat_2(x_cat_2)
+
+        x_cat_3 = torch.cat([x_cat_2, x_fea3], dim=1)
+
+        x_cat_3 = self.b_block_3(x_cat_3)
+        x_cat_3 = self.b_concat_3(x_cat_3)
+
+        x_cat_4 = torch.cat([x_cat_3, x_fea4], dim=1)
+
+        x_cat_4 = self.b_block_4(x_cat_4)
+        x_cat_4 = self.b_concat_4(x_cat_4)
+
+        x_cat_4 = self.b_LR_conv(x_cat_4)
+
+        # short cut
+        x_cat_4 = x_cat_4 + x_b_fea
+        x_branch = self.b_module(x_cat_4)
+
+        # x_out_branch = self.conv_w(x_branch)
+        ########
+        x_branch_d = x_branch
+        x_f_cat = torch.cat([x_branch_d, x], dim=1)
+        x_f_cat = self.f_block(x_f_cat)
+        x_out = self.f_concat(x_f_cat)
+        x_out = self.f_HR_conv0(x_out)
+        x_out = self.f_HR_conv1(x_out)
+
+        #########
+        # return x_out_branch, x_out, x_grad
+        return x_out

backend/headless/fcbh_extras/chainner_models/architecture/SRVGG.py

@@ -0,0 +1,114 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+import math
+
+import torch.nn as nn
+import torch.nn.functional as F
+
+
+class SRVGGNetCompact(nn.Module):
+    """A compact VGG-style network structure for super-resolution.
+    It is a compact network structure, which performs upsampling in the last layer and no convolution is
+    conducted on the HR feature space.
+    Args:
+        num_in_ch (int): Channel number of inputs. Default: 3.
+        num_out_ch (int): Channel number of outputs. Default: 3.
+        num_feat (int): Channel number of intermediate features. Default: 64.
+        num_conv (int): Number of convolution layers in the body network. Default: 16.
+        upscale (int): Upsampling factor. Default: 4.
+        act_type (str): Activation type, options: 'relu', 'prelu', 'leakyrelu'. Default: prelu.
+    """
+
+    def __init__(
+        self,
+        state_dict,
+        act_type: str = "prelu",
+    ):
+        super(SRVGGNetCompact, self).__init__()
+        self.model_arch = "SRVGG (RealESRGAN)"
+        self.sub_type = "SR"
+
+        self.act_type = act_type
+
+        self.state = state_dict
+
+        if "params" in self.state:
+            self.state = self.state["params"]
+
+        self.key_arr = list(self.state.keys())
+
+        self.in_nc = self.get_in_nc()
+        self.num_feat = self.get_num_feats()
+        self.num_conv = self.get_num_conv()
+        self.out_nc = self.in_nc  # :(
+        self.pixelshuffle_shape = None  # Defined in get_scale()
+        self.scale = self.get_scale()
+
+        self.supports_fp16 = True
+        self.supports_bfp16 = True
+        self.min_size_restriction = None
+
+        self.body = nn.ModuleList()
+        # the first conv
+        self.body.append(nn.Conv2d(self.in_nc, self.num_feat, 3, 1, 1))
+        # the first activation
+        if act_type == "relu":
+            activation = nn.ReLU(inplace=True)
+        elif act_type == "prelu":
+            activation = nn.PReLU(num_parameters=self.num_feat)
+        elif act_type == "leakyrelu":
+            activation = nn.LeakyReLU(negative_slope=0.1, inplace=True)
+        self.body.append(activation)  # type: ignore
+
+        # the body structure
+        for _ in range(self.num_conv):
+            self.body.append(nn.Conv2d(self.num_feat, self.num_feat, 3, 1, 1))
+            # activation
+            if act_type == "relu":
+                activation = nn.ReLU(inplace=True)
+            elif act_type == "prelu":
+                activation = nn.PReLU(num_parameters=self.num_feat)
+            elif act_type == "leakyrelu":
+                activation = nn.LeakyReLU(negative_slope=0.1, inplace=True)
+            self.body.append(activation)  # type: ignore
+
+        # the last conv
+        self.body.append(nn.Conv2d(self.num_feat, self.pixelshuffle_shape, 3, 1, 1))  # type: ignore
+        # upsample
+        self.upsampler = nn.PixelShuffle(self.scale)
+
+        self.load_state_dict(self.state, strict=False)
+
+    def get_num_conv(self) -> int:
+        return (int(self.key_arr[-1].split(".")[1]) - 2) // 2
+
+    def get_num_feats(self) -> int:
+        return self.state[self.key_arr[0]].shape[0]
+
+    def get_in_nc(self) -> int:
+        return self.state[self.key_arr[0]].shape[1]
+
+    def get_scale(self) -> int:
+        self.pixelshuffle_shape = self.state[self.key_arr[-1]].shape[0]
+        # Assume out_nc is the same as in_nc
+        # I cant think of a better way to do that
+        self.out_nc = self.in_nc
+        scale = math.sqrt(self.pixelshuffle_shape / self.out_nc)
+        if scale - int(scale) > 0:
+            print(
+                "out_nc is probably different than in_nc, scale calculation might be wrong"
+            )
+        scale = int(scale)
+        return scale
+
+    def forward(self, x):
+        out = x
+        for i in range(0, len(self.body)):
+            out = self.body[i](out)
+
+        out = self.upsampler(out)
+        # add the nearest upsampled image, so that the network learns the residual
+        base = F.interpolate(x, scale_factor=self.scale, mode="nearest")
+        out += base
+        return out

backend/headless/fcbh_extras/chainner_models/architecture/SwiftSRGAN.py

@@ -0,0 +1,161 @@
+# From https://github.com/Koushik0901/Swift-SRGAN/blob/master/swift-srgan/models.py
+
+import torch
+from torch import nn
+
+
+class SeperableConv2d(nn.Module):
+    def __init__(
+        self, in_channels, out_channels, kernel_size, stride=1, padding=1, bias=True
+    ):
+        super(SeperableConv2d, self).__init__()
+        self.depthwise = nn.Conv2d(
+            in_channels,
+            in_channels,
+            kernel_size=kernel_size,
+            stride=stride,
+            groups=in_channels,
+            bias=bias,
+            padding=padding,
+        )
+        self.pointwise = nn.Conv2d(in_channels, out_channels, kernel_size=1, bias=bias)
+
+    def forward(self, x):
+        return self.pointwise(self.depthwise(x))
+
+
+class ConvBlock(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        out_channels,
+        use_act=True,
+        use_bn=True,
+        discriminator=False,
+        **kwargs,
+    ):
+        super(ConvBlock, self).__init__()
+
+        self.use_act = use_act
+        self.cnn = SeperableConv2d(in_channels, out_channels, **kwargs, bias=not use_bn)
+        self.bn = nn.BatchNorm2d(out_channels) if use_bn else nn.Identity()
+        self.act = (
+            nn.LeakyReLU(0.2, inplace=True)
+            if discriminator
+            else nn.PReLU(num_parameters=out_channels)
+        )
+
+    def forward(self, x):
+        return self.act(self.bn(self.cnn(x))) if self.use_act else self.bn(self.cnn(x))
+
+
+class UpsampleBlock(nn.Module):
+    def __init__(self, in_channels, scale_factor):
+        super(UpsampleBlock, self).__init__()
+
+        self.conv = SeperableConv2d(
+            in_channels,
+            in_channels * scale_factor**2,
+            kernel_size=3,
+            stride=1,
+            padding=1,
+        )
+        self.ps = nn.PixelShuffle(
+            scale_factor
+        )  # (in_channels * 4, H, W) -> (in_channels, H*2, W*2)
+        self.act = nn.PReLU(num_parameters=in_channels)
+
+    def forward(self, x):
+        return self.act(self.ps(self.conv(x)))
+
+
+class ResidualBlock(nn.Module):
+    def __init__(self, in_channels):
+        super(ResidualBlock, self).__init__()
+
+        self.block1 = ConvBlock(
+            in_channels, in_channels, kernel_size=3, stride=1, padding=1
+        )
+        self.block2 = ConvBlock(
+            in_channels, in_channels, kernel_size=3, stride=1, padding=1, use_act=False
+        )
+
+    def forward(self, x):
+        out = self.block1(x)
+        out = self.block2(out)
+        return out + x
+
+
+class Generator(nn.Module):
+    """Swift-SRGAN Generator
+    Args:
+        in_channels (int): number of input image channels.
+        num_channels (int): number of hidden channels.
+        num_blocks (int): number of residual blocks.
+        upscale_factor (int): factor to upscale the image [2x, 4x, 8x].
+    Returns:
+        torch.Tensor: super resolution image
+    """
+
+    def __init__(
+        self,
+        state_dict,
+    ):
+        super(Generator, self).__init__()
+        self.model_arch = "Swift-SRGAN"
+        self.sub_type = "SR"
+        self.state = state_dict
+        if "model" in self.state:
+            self.state = self.state["model"]
+
+        self.in_nc: int = self.state["initial.cnn.depthwise.weight"].shape[0]
+        self.out_nc: int = self.state["final_conv.pointwise.weight"].shape[0]
+        self.num_filters: int = self.state["initial.cnn.pointwise.weight"].shape[0]
+        self.num_blocks = len(
+            set([x.split(".")[1] for x in self.state.keys() if "residual" in x])
+        )
+        self.scale: int = 2 ** len(
+            set([x.split(".")[1] for x in self.state.keys() if "upsampler" in x])
+        )
+
+        in_channels = self.in_nc
+        num_channels = self.num_filters
+        num_blocks = self.num_blocks
+        upscale_factor = self.scale
+
+        self.supports_fp16 = True
+        self.supports_bfp16 = True
+        self.min_size_restriction = None
+
+        self.initial = ConvBlock(
+            in_channels, num_channels, kernel_size=9, stride=1, padding=4, use_bn=False
+        )
+        self.residual = nn.Sequential(
+            *[ResidualBlock(num_channels) for _ in range(num_blocks)]
+        )
+        self.convblock = ConvBlock(
+            num_channels,
+            num_channels,
+            kernel_size=3,
+            stride=1,
+            padding=1,
+            use_act=False,
+        )
+        self.upsampler = nn.Sequential(
+            *[
+                UpsampleBlock(num_channels, scale_factor=2)
+                for _ in range(upscale_factor // 2)
+            ]
+        )
+        self.final_conv = SeperableConv2d(
+            num_channels, in_channels, kernel_size=9, stride=1, padding=4
+        )
+
+        self.load_state_dict(self.state, strict=False)
+
+    def forward(self, x):
+        initial = self.initial(x)
+        x = self.residual(initial)
+        x = self.convblock(x) + initial
+        x = self.upsampler(x)
+        return (torch.tanh(self.final_conv(x)) + 1) / 2

backend/headless/fcbh_extras/chainner_models/architecture/block.py

@@ -0,0 +1,546 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+from __future__ import annotations
+
+from collections import OrderedDict
+try:
+    from typing import Literal
+except ImportError:
+    from typing_extensions import Literal
+
+import torch
+import torch.nn as nn
+
+####################
+# Basic blocks
+####################
+
+
+def act(act_type: str, inplace=True, neg_slope=0.2, n_prelu=1):
+    # helper selecting activation
+    # neg_slope: for leakyrelu and init of prelu
+    # n_prelu: for p_relu num_parameters
+    act_type = act_type.lower()
+    if act_type == "relu":
+        layer = nn.ReLU(inplace)
+    elif act_type == "leakyrelu":
+        layer = nn.LeakyReLU(neg_slope, inplace)
+    elif act_type == "prelu":
+        layer = nn.PReLU(num_parameters=n_prelu, init=neg_slope)
+    else:
+        raise NotImplementedError(
+            "activation layer [{:s}] is not found".format(act_type)
+        )
+    return layer
+
+
+def norm(norm_type: str, nc: int):
+    # helper selecting normalization layer
+    norm_type = norm_type.lower()
+    if norm_type == "batch":
+        layer = nn.BatchNorm2d(nc, affine=True)
+    elif norm_type == "instance":
+        layer = nn.InstanceNorm2d(nc, affine=False)
+    else:
+        raise NotImplementedError(
+            "normalization layer [{:s}] is not found".format(norm_type)
+        )
+    return layer
+
+
+def pad(pad_type: str, padding):
+    # helper selecting padding layer
+    # if padding is 'zero', do by conv layers
+    pad_type = pad_type.lower()
+    if padding == 0:
+        return None
+    if pad_type == "reflect":
+        layer = nn.ReflectionPad2d(padding)
+    elif pad_type == "replicate":
+        layer = nn.ReplicationPad2d(padding)
+    else:
+        raise NotImplementedError(
+            "padding layer [{:s}] is not implemented".format(pad_type)
+        )
+    return layer
+
+
+def get_valid_padding(kernel_size, dilation):
+    kernel_size = kernel_size + (kernel_size - 1) * (dilation - 1)
+    padding = (kernel_size - 1) // 2
+    return padding
+
+
+class ConcatBlock(nn.Module):
+    # Concat the output of a submodule to its input
+    def __init__(self, submodule):
+        super(ConcatBlock, self).__init__()
+        self.sub = submodule
+
+    def forward(self, x):
+        output = torch.cat((x, self.sub(x)), dim=1)
+        return output
+
+    def __repr__(self):
+        tmpstr = "Identity .. \n|"
+        modstr = self.sub.__repr__().replace("\n", "\n|")
+        tmpstr = tmpstr + modstr
+        return tmpstr
+
+
+class ShortcutBlock(nn.Module):
+    # Elementwise sum the output of a submodule to its input
+    def __init__(self, submodule):
+        super(ShortcutBlock, self).__init__()
+        self.sub = submodule
+
+    def forward(self, x):
+        output = x + self.sub(x)
+        return output
+
+    def __repr__(self):
+        tmpstr = "Identity + \n|"
+        modstr = self.sub.__repr__().replace("\n", "\n|")
+        tmpstr = tmpstr + modstr
+        return tmpstr
+
+
+class ShortcutBlockSPSR(nn.Module):
+    # Elementwise sum the output of a submodule to its input
+    def __init__(self, submodule):
+        super(ShortcutBlockSPSR, self).__init__()
+        self.sub = submodule
+
+    def forward(self, x):
+        return x, self.sub
+
+    def __repr__(self):
+        tmpstr = "Identity + \n|"
+        modstr = self.sub.__repr__().replace("\n", "\n|")
+        tmpstr = tmpstr + modstr
+        return tmpstr
+
+
+def sequential(*args):
+    # Flatten Sequential. It unwraps nn.Sequential.
+    if len(args) == 1:
+        if isinstance(args[0], OrderedDict):
+            raise NotImplementedError("sequential does not support OrderedDict input.")
+        return args[0]  # No sequential is needed.
+    modules = []
+    for module in args:
+        if isinstance(module, nn.Sequential):
+            for submodule in module.children():
+                modules.append(submodule)
+        elif isinstance(module, nn.Module):
+            modules.append(module)
+    return nn.Sequential(*modules)
+
+
+ConvMode = Literal["CNA", "NAC", "CNAC"]
+
+
+# 2x2x2 Conv Block
+def conv_block_2c2(
+    in_nc,
+    out_nc,
+    act_type="relu",
+):
+    return sequential(
+        nn.Conv2d(in_nc, out_nc, kernel_size=2, padding=1),
+        nn.Conv2d(out_nc, out_nc, kernel_size=2, padding=0),
+        act(act_type) if act_type else None,
+    )
+
+
+def conv_block(
+    in_nc: int,
+    out_nc: int,
+    kernel_size,
+    stride=1,
+    dilation=1,
+    groups=1,
+    bias=True,
+    pad_type="zero",
+    norm_type: str | None = None,
+    act_type: str | None = "relu",
+    mode: ConvMode = "CNA",
+    c2x2=False,
+):
+    """
+    Conv layer with padding, normalization, activation
+    mode: CNA --> Conv -> Norm -> Act
+        NAC --> Norm -> Act --> Conv (Identity Mappings in Deep Residual Networks, ECCV16)
+    """
+
+    if c2x2:
+        return conv_block_2c2(in_nc, out_nc, act_type=act_type)
+
+    assert mode in ("CNA", "NAC", "CNAC"), "Wrong conv mode [{:s}]".format(mode)
+    padding = get_valid_padding(kernel_size, dilation)
+    p = pad(pad_type, padding) if pad_type and pad_type != "zero" else None
+    padding = padding if pad_type == "zero" else 0
+
+    c = nn.Conv2d(
+        in_nc,
+        out_nc,
+        kernel_size=kernel_size,
+        stride=stride,
+        padding=padding,
+        dilation=dilation,
+        bias=bias,
+        groups=groups,
+    )
+    a = act(act_type) if act_type else None
+    if mode in ("CNA", "CNAC"):
+        n = norm(norm_type, out_nc) if norm_type else None
+        return sequential(p, c, n, a)
+    elif mode == "NAC":
+        if norm_type is None and act_type is not None:
+            a = act(act_type, inplace=False)
+            # Important!
+            # input----ReLU(inplace)----Conv--+----output
+            #        |________________________|
+            # inplace ReLU will modify the input, therefore wrong output
+        n = norm(norm_type, in_nc) if norm_type else None
+        return sequential(n, a, p, c)
+    else:
+        assert False, f"Invalid conv mode {mode}"
+
+
+####################
+# Useful blocks
+####################
+
+
+class ResNetBlock(nn.Module):
+    """
+    ResNet Block, 3-3 style
+    with extra residual scaling used in EDSR
+    (Enhanced Deep Residual Networks for Single Image Super-Resolution, CVPRW 17)
+    """
+
+    def __init__(
+        self,
+        in_nc,
+        mid_nc,
+        out_nc,
+        kernel_size=3,
+        stride=1,
+        dilation=1,
+        groups=1,
+        bias=True,
+        pad_type="zero",
+        norm_type=None,
+        act_type="relu",
+        mode: ConvMode = "CNA",
+        res_scale=1,
+    ):
+        super(ResNetBlock, self).__init__()
+        conv0 = conv_block(
+            in_nc,
+            mid_nc,
+            kernel_size,
+            stride,
+            dilation,
+            groups,
+            bias,
+            pad_type,
+            norm_type,
+            act_type,
+            mode,
+        )
+        if mode == "CNA":
+            act_type = None
+        if mode == "CNAC":  # Residual path: |-CNAC-|
+            act_type = None
+            norm_type = None
+        conv1 = conv_block(
+            mid_nc,
+            out_nc,
+            kernel_size,
+            stride,
+            dilation,
+            groups,
+            bias,
+            pad_type,
+            norm_type,
+            act_type,
+            mode,
+        )
+        # if in_nc != out_nc:
+        #     self.project = conv_block(in_nc, out_nc, 1, stride, dilation, 1, bias, pad_type, \
+        #         None, None)
+        #     print('Need a projecter in ResNetBlock.')
+        # else:
+        #     self.project = lambda x:x
+        self.res = sequential(conv0, conv1)
+        self.res_scale = res_scale
+
+    def forward(self, x):
+        res = self.res(x).mul(self.res_scale)
+        return x + res
+
+
+class RRDB(nn.Module):
+    """
+    Residual in Residual Dense Block
+    (ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks)
+    """
+
+    def __init__(
+        self,
+        nf,
+        kernel_size=3,
+        gc=32,
+        stride=1,
+        bias: bool = True,
+        pad_type="zero",
+        norm_type=None,
+        act_type="leakyrelu",
+        mode: ConvMode = "CNA",
+        _convtype="Conv2D",
+        _spectral_norm=False,
+        plus=False,
+        c2x2=False,
+    ):
+        super(RRDB, self).__init__()
+        self.RDB1 = ResidualDenseBlock_5C(
+            nf,
+            kernel_size,
+            gc,
+            stride,
+            bias,
+            pad_type,
+            norm_type,
+            act_type,
+            mode,
+            plus=plus,
+            c2x2=c2x2,
+        )
+        self.RDB2 = ResidualDenseBlock_5C(
+            nf,
+            kernel_size,
+            gc,
+            stride,
+            bias,
+            pad_type,
+            norm_type,
+            act_type,
+            mode,
+            plus=plus,
+            c2x2=c2x2,
+        )
+        self.RDB3 = ResidualDenseBlock_5C(
+            nf,
+            kernel_size,
+            gc,
+            stride,
+            bias,
+            pad_type,
+            norm_type,
+            act_type,
+            mode,
+            plus=plus,
+            c2x2=c2x2,
+        )
+
+    def forward(self, x):
+        out = self.RDB1(x)
+        out = self.RDB2(out)
+        out = self.RDB3(out)
+        return out * 0.2 + x
+
+
+class ResidualDenseBlock_5C(nn.Module):
+    """
+    Residual Dense Block
+    style: 5 convs
+    The core module of paper: (Residual Dense Network for Image Super-Resolution, CVPR 18)
+    Modified options that can be used:
+        - "Partial Convolution based Padding" arXiv:1811.11718
+        - "Spectral normalization" arXiv:1802.05957
+        - "ICASSP 2020 - ESRGAN+ : Further Improving ESRGAN" N. C.
+            {Rakotonirina} and A. {Rasoanaivo}
+
+    Args:
+        nf (int): Channel number of intermediate features (num_feat).
+        gc (int): Channels for each growth (num_grow_ch: growth channel,
+            i.e. intermediate channels).
+        convtype (str): the type of convolution to use. Default: 'Conv2D'
+        gaussian_noise (bool): enable the ESRGAN+ gaussian noise (no new
+            trainable parameters)
+        plus (bool): enable the additional residual paths from ESRGAN+
+            (adds trainable parameters)
+    """
+
+    def __init__(
+        self,
+        nf=64,
+        kernel_size=3,
+        gc=32,
+        stride=1,
+        bias: bool = True,
+        pad_type="zero",
+        norm_type=None,
+        act_type="leakyrelu",
+        mode: ConvMode = "CNA",
+        plus=False,
+        c2x2=False,
+    ):
+        super(ResidualDenseBlock_5C, self).__init__()
+
+        ## +
+        self.conv1x1 = conv1x1(nf, gc) if plus else None
+        ## +
+
+        self.conv1 = conv_block(
+            nf,
+            gc,
+            kernel_size,
+            stride,
+            bias=bias,
+            pad_type=pad_type,
+            norm_type=norm_type,
+            act_type=act_type,
+            mode=mode,
+            c2x2=c2x2,
+        )
+        self.conv2 = conv_block(
+            nf + gc,
+            gc,
+            kernel_size,
+            stride,
+            bias=bias,
+            pad_type=pad_type,
+            norm_type=norm_type,
+            act_type=act_type,
+            mode=mode,
+            c2x2=c2x2,
+        )
+        self.conv3 = conv_block(
+            nf + 2 * gc,
+            gc,
+            kernel_size,
+            stride,
+            bias=bias,
+            pad_type=pad_type,
+            norm_type=norm_type,
+            act_type=act_type,
+            mode=mode,
+            c2x2=c2x2,
+        )
+        self.conv4 = conv_block(
+            nf + 3 * gc,
+            gc,
+            kernel_size,
+            stride,
+            bias=bias,
+            pad_type=pad_type,
+            norm_type=norm_type,
+            act_type=act_type,
+            mode=mode,
+            c2x2=c2x2,
+        )
+        if mode == "CNA":
+            last_act = None
+        else:
+            last_act = act_type
+        self.conv5 = conv_block(
+            nf + 4 * gc,
+            nf,
+            3,
+            stride,
+            bias=bias,
+            pad_type=pad_type,
+            norm_type=norm_type,
+            act_type=last_act,
+            mode=mode,
+            c2x2=c2x2,
+        )
+
+    def forward(self, x):
+        x1 = self.conv1(x)
+        x2 = self.conv2(torch.cat((x, x1), 1))
+        if self.conv1x1:
+            # pylint: disable=not-callable
+            x2 = x2 + self.conv1x1(x)  # +
+        x3 = self.conv3(torch.cat((x, x1, x2), 1))
+        x4 = self.conv4(torch.cat((x, x1, x2, x3), 1))
+        if self.conv1x1:
+            x4 = x4 + x2  # +
+        x5 = self.conv5(torch.cat((x, x1, x2, x3, x4), 1))
+        return x5 * 0.2 + x
+
+
+def conv1x1(in_planes, out_planes, stride=1):
+    return nn.Conv2d(in_planes, out_planes, kernel_size=1, stride=stride, bias=False)
+
+
+####################
+# Upsampler
+####################
+
+
+def pixelshuffle_block(
+    in_nc: int,
+    out_nc: int,
+    upscale_factor=2,
+    kernel_size=3,
+    stride=1,
+    bias=True,
+    pad_type="zero",
+    norm_type: str | None = None,
+    act_type="relu",
+):
+    """
+    Pixel shuffle layer
+    (Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional
+    Neural Network, CVPR17)
+    """
+    conv = conv_block(
+        in_nc,
+        out_nc * (upscale_factor**2),
+        kernel_size,
+        stride,
+        bias=bias,
+        pad_type=pad_type,
+        norm_type=None,
+        act_type=None,
+    )
+    pixel_shuffle = nn.PixelShuffle(upscale_factor)
+
+    n = norm(norm_type, out_nc) if norm_type else None
+    a = act(act_type) if act_type else None
+    return sequential(conv, pixel_shuffle, n, a)
+
+
+def upconv_block(
+    in_nc: int,
+    out_nc: int,
+    upscale_factor=2,
+    kernel_size=3,
+    stride=1,
+    bias=True,
+    pad_type="zero",
+    norm_type: str | None = None,
+    act_type="relu",
+    mode="nearest",
+    c2x2=False,
+):
+    # Up conv
+    # described in https://distill.pub/2016/deconv-checkerboard/
+    upsample = nn.Upsample(scale_factor=upscale_factor, mode=mode)
+    conv = conv_block(
+        in_nc,
+        out_nc,
+        kernel_size,
+        stride,
+        bias=bias,
+        pad_type=pad_type,
+        norm_type=norm_type,
+        act_type=act_type,
+        c2x2=c2x2,
+    )
+    return sequential(upsample, conv)

backend/headless/fcbh_extras/chainner_models/architecture/face/LICENSE-GFPGAN

@@ -0,0 +1,351 @@
+Tencent is pleased to support the open source community by making GFPGAN available.
+
+Copyright (C) 2021 THL A29 Limited, a Tencent company.  All rights reserved.
+
+GFPGAN is licensed under the Apache License Version 2.0 except for the third-party components listed below.
+
+
+Terms of the Apache License Version 2.0:
+---------------------------------------------
+Apache License
+
+Version 2.0, January 2004
+
+http://www.apache.org/licenses/
+
+TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+1. Definitions.
+
+“License” shall mean the terms and conditions for use, reproduction, and distribution as defined by Sections 1 through 9 of this document.
+
+“Licensor” shall mean the copyright owner or entity authorized by the copyright owner that is granting the License.
+
+“Legal Entity” shall mean the union of the acting entity and all other entities that control, are controlled by, or are under common control with that entity. For the purposes of this definition, “control” means (i) the power, direct or indirect, to cause the direction or management of such entity, whether by contract or otherwise, or (ii) ownership of fifty percent (50%) or more of the outstanding shares, or (iii) beneficial ownership of such entity.
+
+“You” (or “Your”) shall mean an individual or Legal Entity exercising permissions granted by this License.
+
+“Source” form shall mean the preferred form for making modifications, including but not limited to software source code, documentation source, and configuration files.
+
+“Object” form shall mean any form resulting from mechanical transformation or translation of a Source form, including but not limited to compiled object code, generated documentation, and conversions to other media types.
+
+“Work” shall mean the work of authorship, whether in Source or Object form, made available under the License, as indicated by a copyright notice that is included in or attached to the work (an example is provided in the Appendix below).
+
+“Derivative Works” shall mean any work, whether in Source or Object form, that is based on (or derived from) the Work and for which the editorial revisions, annotations, elaborations, or other modifications represent, as a whole, an original work of authorship. For the purposes of this License, Derivative Works shall not include works that remain separable from, or merely link (or bind by name) to the interfaces of, the Work and Derivative Works thereof.
+
+“Contribution” shall mean any work of authorship, including the original version of the Work and any modifications or additions to that Work or Derivative Works thereof, that is intentionally submitted to Licensor for inclusion in the Work by the copyright owner or by an individual or Legal Entity authorized to submit on behalf of the copyright owner. For the purposes of this definition, “submitted” means any form of electronic, verbal, or written communication sent to the Licensor or its representatives, including but not limited to communication on electronic mailing lists, source code control systems, and issue tracking systems that are managed by, or on behalf of, the Licensor for the purpose of discussing and improving the Work, but excluding communication that is conspicuously marked or otherwise designated in writing by the copyright owner as “Not a Contribution.”
+
+“Contributor” shall mean Licensor and any individual or Legal Entity on behalf of whom a Contribution has been received by Licensor and subsequently incorporated within the Work.
+
+2. Grant of Copyright License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable copyright license to reproduce, prepare Derivative Works of, publicly display, publicly perform, sublicense, and distribute the Work and such Derivative Works in Source or Object form.
+
+3. Grant of Patent License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable (except as stated in this section) patent license to make, have made, use, offer to sell, sell, import, and otherwise transfer the Work, where such license applies only to those patent claims licensable by such Contributor that are necessarily infringed by their Contribution(s) alone or by combination of their Contribution(s) with the Work to which such Contribution(s) was submitted. If You institute patent litigation against any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Work or a Contribution incorporated within the Work constitutes direct or contributory patent infringement, then any patent licenses granted to You under this License for that Work shall terminate as of the date such litigation is filed.
+
+4. Redistribution. You may reproduce and distribute copies of the Work or Derivative Works thereof in any medium, with or without modifications, and in Source or Object form, provided that You meet the following conditions:
+
+You must give any other recipients of the Work or Derivative Works a copy of this License; and
+
+You must cause any modified files to carry prominent notices stating that You changed the files; and
+
+You must retain, in the Source form of any Derivative Works that You distribute, all copyright, patent, trademark, and attribution notices from the Source form of the Work, excluding those notices that do not pertain to any part of the Derivative Works; and
+
+If the Work includes a “NOTICE” text file as part of its distribution, then any Derivative Works that You distribute must include a readable copy of the attribution notices contained within such NOTICE file, excluding those notices that do not pertain to any part of the Derivative Works, in at least one of the following places: within a NOTICE text file distributed as part of the Derivative Works; within the Source form or documentation, if provided along with the Derivative Works; or, within a display generated by the Derivative Works, if and wherever such third-party notices normally appear. The contents of the NOTICE file are for informational purposes only and do not modify the License. You may add Your own attribution notices within Derivative Works that You distribute, alongside or as an addendum to the NOTICE text from the Work, provided that such additional attribution notices cannot be construed as modifying the License.
+
+You may add Your own copyright statement to Your modifications and may provide additional or different license terms and conditions for use, reproduction, or distribution of Your modifications, or for any such Derivative Works as a whole, provided Your use, reproduction, and distribution of the Work otherwise complies with the conditions stated in this License.
+
+5. Submission of Contributions. Unless You explicitly state otherwise, any Contribution intentionally submitted for inclusion in the Work by You to the Licensor shall be under the terms and conditions of this License, without any additional terms or conditions. Notwithstanding the above, nothing herein shall supersede or modify the terms of any separate license agreement you may have executed with Licensor regarding such Contributions.
+
+6. Trademarks. This License does not grant permission to use the trade names, trademarks, service marks, or product names of the Licensor, except as required for reasonable and customary use in describing the origin of the Work and reproducing the content of the NOTICE file.
+
+7. Disclaimer of Warranty. Unless required by applicable law or agreed to in writing, Licensor provides the Work (and each Contributor provides its Contributions) on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied, including, without limitation, any warranties or conditions of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A PARTICULAR PURPOSE. You are solely responsible for determining the appropriateness of using or redistributing the Work and assume any risks associated with Your exercise of permissions under this License.
+
+8. Limitation of Liability. In no event and under no legal theory, whether in tort (including negligence), contract, or otherwise, unless required by applicable law (such as deliberate and grossly negligent acts) or agreed to in writing, shall any Contributor be liable to You for damages, including any direct, indirect, special, incidental, or consequential damages of any character arising as a result of this License or out of the use or inability to use the Work (including but not limited to damages for loss of goodwill, work stoppage, computer failure or malfunction, or any and all other commercial damages or losses), even if such Contributor has been advised of the possibility of such damages.
+
+9. Accepting Warranty or Additional Liability. While redistributing the Work or Derivative Works thereof, You may choose to offer, and charge a fee for, acceptance of support, warranty, indemnity, or other liability obligations and/or rights consistent with this License. However, in accepting such obligations, You may act only on Your own behalf and on Your sole responsibility, not on behalf of any other Contributor, and only if You agree to indemnify, defend, and hold each Contributor harmless for any liability incurred by, or claims asserted against, such Contributor by reason of your accepting any such warranty or additional liability.
+
+END OF TERMS AND CONDITIONS
+
+
+
+Other  dependencies and licenses:
+
+
+Open Source Software licensed under the Apache 2.0 license and Other Licenses of the Third-Party Components therein:
+---------------------------------------------
+1. basicsr
+Copyright 2018-2020 BasicSR Authors
+
+
+This BasicSR project is released under the Apache 2.0 license.
+
+A copy of Apache 2.0 is included in this file.
+
+StyleGAN2
+The codes are modified from the repository stylegan2-pytorch. Many thanks to the author - Kim Seonghyeon 😊 for translating from the official TensorFlow codes to PyTorch ones. Here is the license of stylegan2-pytorch.
+The official repository is https://github.com/NVlabs/stylegan2, and here is the NVIDIA license.
+DFDNet
+The codes are largely modified from the repository DFDNet. Their license is Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
+
+Terms of the Nvidia License:
+---------------------------------------------
+
+1. Definitions
+
+"Licensor" means any person or entity that distributes its Work.
+
+"Software" means the original work of authorship made available under
+this License.
+
+"Work" means the Software and any additions to or derivative works of
+the Software that are made available under this License.
+
+"Nvidia Processors" means any central processing unit (CPU), graphics
+processing unit (GPU), field-programmable gate array (FPGA),
+application-specific integrated circuit (ASIC) or any combination
+thereof designed, made, sold, or provided by Nvidia or its affiliates.
+
+The terms "reproduce," "reproduction," "derivative works," and
+"distribution" have the meaning as provided under U.S. copyright law;
+provided, however, that for the purposes of this License, derivative
+works shall not include works that remain separable from, or merely
+link (or bind by name) to the interfaces of, the Work.
+
+Works, including the Software, are "made available" under this License
+by including in or with the Work either (a) a copyright notice
+referencing the applicability of this License to the Work, or (b) a
+copy of this License.
+
+2. License Grants
+
+    2.1 Copyright Grant. Subject to the terms and conditions of this
+    License, each Licensor grants to you a perpetual, worldwide,
+    non-exclusive, royalty-free, copyright license to reproduce,
+    prepare derivative works of, publicly display, publicly perform,
+    sublicense and distribute its Work and any resulting derivative
+    works in any form.
+
+3. Limitations
+
+    3.1 Redistribution. You may reproduce or distribute the Work only
+    if (a) you do so under this License, (b) you include a complete
+    copy of this License with your distribution, and (c) you retain
+    without modification any copyright, patent, trademark, or
+    attribution notices that are present in the Work.
+
+    3.2 Derivative Works. You may specify that additional or different
+    terms apply to the use, reproduction, and distribution of your
+    derivative works of the Work ("Your Terms") only if (a) Your Terms
+    provide that the use limitation in Section 3.3 applies to your
+    derivative works, and (b) you identify the specific derivative
+    works that are subject to Your Terms. Notwithstanding Your Terms,
+    this License (including the redistribution requirements in Section
+    3.1) will continue to apply to the Work itself.
+
+    3.3 Use Limitation. The Work and any derivative works thereof only
+    may be used or intended for use non-commercially. The Work or
+    derivative works thereof may be used or intended for use by Nvidia
+    or its affiliates commercially or non-commercially. As used herein,
+    "non-commercially" means for research or evaluation purposes only.
+
+    3.4 Patent Claims. If you bring or threaten to bring a patent claim
+    against any Licensor (including any claim, cross-claim or
+    counterclaim in a lawsuit) to enforce any patents that you allege
+    are infringed by any Work, then your rights under this License from
+    such Licensor (including the grants in Sections 2.1 and 2.2) will
+    terminate immediately.
+
+    3.5 Trademarks. This License does not grant any rights to use any
+    Licensor's or its affiliates' names, logos, or trademarks, except
+    as necessary to reproduce the notices described in this License.
+
+    3.6 Termination. If you violate any term of this License, then your
+    rights under this License (including the grants in Sections 2.1 and
+    2.2) will terminate immediately.
+
+4. Disclaimer of Warranty.
+
+THE WORK IS PROVIDED "AS IS" WITHOUT WARRANTIES OR CONDITIONS OF ANY
+KIND, EITHER EXPRESS OR IMPLIED, INCLUDING WARRANTIES OR CONDITIONS OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, TITLE OR
+NON-INFRINGEMENT. YOU BEAR THE RISK OF UNDERTAKING ANY ACTIVITIES UNDER
+THIS LICENSE.
+
+5. Limitation of Liability.
+
+EXCEPT AS PROHIBITED BY APPLICABLE LAW, IN NO EVENT AND UNDER NO LEGAL
+THEORY, WHETHER IN TORT (INCLUDING NEGLIGENCE), CONTRACT, OR OTHERWISE
+SHALL ANY LICENSOR BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY DIRECT,
+INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING OUT OF
+OR RELATED TO THIS LICENSE, THE USE OR INABILITY TO USE THE WORK
+(INCLUDING BUT NOT LIMITED TO LOSS OF GOODWILL, BUSINESS INTERRUPTION,
+LOST PROFITS OR DATA, COMPUTER FAILURE OR MALFUNCTION, OR ANY OTHER
+COMMERCIAL DAMAGES OR LOSSES), EVEN IF THE LICENSOR HAS BEEN ADVISED OF
+THE POSSIBILITY OF SUCH DAMAGES.
+
+MIT License
+
+Copyright (c) 2019 Kim Seonghyeon
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+
+
+
+Open Source Software licensed under the BSD 3-Clause license:
+---------------------------------------------
+1. torchvision
+Copyright (c) Soumith Chintala 2016,
+All rights reserved.
+
+2. torch
+Copyright (c) 2016-     Facebook, Inc            (Adam Paszke)
+Copyright (c) 2014-     Facebook, Inc            (Soumith Chintala)
+Copyright (c) 2011-2014 Idiap Research Institute (Ronan Collobert)
+Copyright (c) 2012-2014 Deepmind Technologies    (Koray Kavukcuoglu)
+Copyright (c) 2011-2012 NEC Laboratories America (Koray Kavukcuoglu)
+Copyright (c) 2011-2013 NYU                      (Clement Farabet)
+Copyright (c) 2006-2010 NEC Laboratories America (Ronan Collobert, Leon Bottou, Iain Melvin, Jason Weston)
+Copyright (c) 2006      Idiap Research Institute (Samy Bengio)
+Copyright (c) 2001-2004 Idiap Research Institute (Ronan Collobert, Samy Bengio, Johnny Mariethoz)
+
+
+Terms of the BSD 3-Clause License:
+---------------------------------------------
+Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
+
+1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
+
+3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+
+
+Open Source Software licensed under the BSD 3-Clause License and Other Licenses of the Third-Party Components therein:
+---------------------------------------------
+1. numpy
+Copyright (c) 2005-2020, NumPy Developers.
+All rights reserved.
+
+A copy of BSD 3-Clause License is included in this file.
+
+The NumPy repository and source distributions bundle several libraries that are
+compatibly licensed.  We list these here.
+
+Name: Numpydoc
+Files: doc/sphinxext/numpydoc/*
+License: BSD-2-Clause
+  For details, see doc/sphinxext/LICENSE.txt
+
+Name: scipy-sphinx-theme
+Files: doc/scipy-sphinx-theme/*
+License: BSD-3-Clause AND PSF-2.0 AND Apache-2.0
+  For details, see doc/scipy-sphinx-theme/LICENSE.txt
+
+Name: lapack-lite
+Files: numpy/linalg/lapack_lite/*
+License: BSD-3-Clause
+  For details, see numpy/linalg/lapack_lite/LICENSE.txt
+
+Name: tempita
+Files: tools/npy_tempita/*
+License: MIT
+  For details, see tools/npy_tempita/license.txt
+
+Name: dragon4
+Files: numpy/core/src/multiarray/dragon4.c
+License: MIT
+  For license text, see numpy/core/src/multiarray/dragon4.c
+
+
+
+Open Source Software licensed under the MIT license:
+---------------------------------------------
+1. facexlib
+Copyright (c) 2020 Xintao Wang
+
+2. opencv-python
+Copyright (c) Olli-Pekka Heinisuo
+Please note that only files in cv2 package are used.
+
+
+Terms of the MIT License:
+---------------------------------------------
+Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+
+
+Open Source Software licensed under the MIT license and Other Licenses of the Third-Party Components therein:
+---------------------------------------------
+1. tqdm
+Copyright (c) 2013 noamraph
+
+`tqdm` is a product of collaborative work.
+Unless otherwise stated, all authors (see commit logs) retain copyright
+for their respective work, and release the work under the MIT licence
+(text below).
+
+Exceptions or notable authors are listed below
+in reverse chronological order:
+
+* files: *
+  MPLv2.0 2015-2020 (c) Casper da Costa-Luis
+  [casperdcl](https://github.com/casperdcl).
+* files: tqdm/_tqdm.py
+  MIT 2016 (c) [PR #96] on behalf of Google Inc.
+* files: tqdm/_tqdm.py setup.py README.rst MANIFEST.in .gitignore
+  MIT 2013 (c) Noam Yorav-Raphael, original author.
+
+[PR #96]: https://github.com/tqdm/tqdm/pull/96
+
+
+Mozilla Public Licence (MPL) v. 2.0 - Exhibit A
+-----------------------------------------------
+
+This Source Code Form is subject to the terms of the
+Mozilla Public License, v. 2.0.
+If a copy of the MPL was not distributed with this file,
+You can obtain one at https://mozilla.org/MPL/2.0/.
+
+
+MIT License (MIT)
+-----------------
+
+Copyright (c) 2013 noamraph
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of
+this software and associated documentation files (the "Software"), to deal in
+the Software without restriction, including without limitation the rights to
+use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of
+the Software, and to permit persons to whom the Software is furnished to do so,
+subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
+FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
+COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
+IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

backend/headless/fcbh_extras/chainner_models/architecture/face/LICENSE-RestoreFormer

@@ -0,0 +1,351 @@
+Tencent is pleased to support the open source community by making GFPGAN available.
+
+Copyright (C) 2021 THL A29 Limited, a Tencent company.  All rights reserved.
+
+GFPGAN is licensed under the Apache License Version 2.0 except for the third-party components listed below.
+
+
+Terms of the Apache License Version 2.0:
+---------------------------------------------
+Apache License
+
+Version 2.0, January 2004
+
+http://www.apache.org/licenses/
+
+TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+1. Definitions.
+
+“License” shall mean the terms and conditions for use, reproduction, and distribution as defined by Sections 1 through 9 of this document.
+
+“Licensor” shall mean the copyright owner or entity authorized by the copyright owner that is granting the License.
+
+“Legal Entity” shall mean the union of the acting entity and all other entities that control, are controlled by, or are under common control with that entity. For the purposes of this definition, “control” means (i) the power, direct or indirect, to cause the direction or management of such entity, whether by contract or otherwise, or (ii) ownership of fifty percent (50%) or more of the outstanding shares, or (iii) beneficial ownership of such entity.
+
+“You” (or “Your”) shall mean an individual or Legal Entity exercising permissions granted by this License.
+
+“Source” form shall mean the preferred form for making modifications, including but not limited to software source code, documentation source, and configuration files.
+
+“Object” form shall mean any form resulting from mechanical transformation or translation of a Source form, including but not limited to compiled object code, generated documentation, and conversions to other media types.
+
+“Work” shall mean the work of authorship, whether in Source or Object form, made available under the License, as indicated by a copyright notice that is included in or attached to the work (an example is provided in the Appendix below).
+
+“Derivative Works” shall mean any work, whether in Source or Object form, that is based on (or derived from) the Work and for which the editorial revisions, annotations, elaborations, or other modifications represent, as a whole, an original work of authorship. For the purposes of this License, Derivative Works shall not include works that remain separable from, or merely link (or bind by name) to the interfaces of, the Work and Derivative Works thereof.
+
+“Contribution” shall mean any work of authorship, including the original version of the Work and any modifications or additions to that Work or Derivative Works thereof, that is intentionally submitted to Licensor for inclusion in the Work by the copyright owner or by an individual or Legal Entity authorized to submit on behalf of the copyright owner. For the purposes of this definition, “submitted” means any form of electronic, verbal, or written communication sent to the Licensor or its representatives, including but not limited to communication on electronic mailing lists, source code control systems, and issue tracking systems that are managed by, or on behalf of, the Licensor for the purpose of discussing and improving the Work, but excluding communication that is conspicuously marked or otherwise designated in writing by the copyright owner as “Not a Contribution.”
+
+“Contributor” shall mean Licensor and any individual or Legal Entity on behalf of whom a Contribution has been received by Licensor and subsequently incorporated within the Work.
+
+2. Grant of Copyright License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable copyright license to reproduce, prepare Derivative Works of, publicly display, publicly perform, sublicense, and distribute the Work and such Derivative Works in Source or Object form.
+
+3. Grant of Patent License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable (except as stated in this section) patent license to make, have made, use, offer to sell, sell, import, and otherwise transfer the Work, where such license applies only to those patent claims licensable by such Contributor that are necessarily infringed by their Contribution(s) alone or by combination of their Contribution(s) with the Work to which such Contribution(s) was submitted. If You institute patent litigation against any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Work or a Contribution incorporated within the Work constitutes direct or contributory patent infringement, then any patent licenses granted to You under this License for that Work shall terminate as of the date such litigation is filed.
+
+4. Redistribution. You may reproduce and distribute copies of the Work or Derivative Works thereof in any medium, with or without modifications, and in Source or Object form, provided that You meet the following conditions:
+
+You must give any other recipients of the Work or Derivative Works a copy of this License; and
+
+You must cause any modified files to carry prominent notices stating that You changed the files; and
+
+You must retain, in the Source form of any Derivative Works that You distribute, all copyright, patent, trademark, and attribution notices from the Source form of the Work, excluding those notices that do not pertain to any part of the Derivative Works; and
+
+If the Work includes a “NOTICE” text file as part of its distribution, then any Derivative Works that You distribute must include a readable copy of the attribution notices contained within such NOTICE file, excluding those notices that do not pertain to any part of the Derivative Works, in at least one of the following places: within a NOTICE text file distributed as part of the Derivative Works; within the Source form or documentation, if provided along with the Derivative Works; or, within a display generated by the Derivative Works, if and wherever such third-party notices normally appear. The contents of the NOTICE file are for informational purposes only and do not modify the License. You may add Your own attribution notices within Derivative Works that You distribute, alongside or as an addendum to the NOTICE text from the Work, provided that such additional attribution notices cannot be construed as modifying the License.
+
+You may add Your own copyright statement to Your modifications and may provide additional or different license terms and conditions for use, reproduction, or distribution of Your modifications, or for any such Derivative Works as a whole, provided Your use, reproduction, and distribution of the Work otherwise complies with the conditions stated in this License.
+
+5. Submission of Contributions. Unless You explicitly state otherwise, any Contribution intentionally submitted for inclusion in the Work by You to the Licensor shall be under the terms and conditions of this License, without any additional terms or conditions. Notwithstanding the above, nothing herein shall supersede or modify the terms of any separate license agreement you may have executed with Licensor regarding such Contributions.
+
+6. Trademarks. This License does not grant permission to use the trade names, trademarks, service marks, or product names of the Licensor, except as required for reasonable and customary use in describing the origin of the Work and reproducing the content of the NOTICE file.
+
+7. Disclaimer of Warranty. Unless required by applicable law or agreed to in writing, Licensor provides the Work (and each Contributor provides its Contributions) on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied, including, without limitation, any warranties or conditions of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A PARTICULAR PURPOSE. You are solely responsible for determining the appropriateness of using or redistributing the Work and assume any risks associated with Your exercise of permissions under this License.
+
+8. Limitation of Liability. In no event and under no legal theory, whether in tort (including negligence), contract, or otherwise, unless required by applicable law (such as deliberate and grossly negligent acts) or agreed to in writing, shall any Contributor be liable to You for damages, including any direct, indirect, special, incidental, or consequential damages of any character arising as a result of this License or out of the use or inability to use the Work (including but not limited to damages for loss of goodwill, work stoppage, computer failure or malfunction, or any and all other commercial damages or losses), even if such Contributor has been advised of the possibility of such damages.
+
+9. Accepting Warranty or Additional Liability. While redistributing the Work or Derivative Works thereof, You may choose to offer, and charge a fee for, acceptance of support, warranty, indemnity, or other liability obligations and/or rights consistent with this License. However, in accepting such obligations, You may act only on Your own behalf and on Your sole responsibility, not on behalf of any other Contributor, and only if You agree to indemnify, defend, and hold each Contributor harmless for any liability incurred by, or claims asserted against, such Contributor by reason of your accepting any such warranty or additional liability.
+
+END OF TERMS AND CONDITIONS
+
+
+
+Other  dependencies and licenses:
+
+
+Open Source Software licensed under the Apache 2.0 license and Other Licenses of the Third-Party Components therein:
+---------------------------------------------
+1. basicsr
+Copyright 2018-2020 BasicSR Authors
+
+
+This BasicSR project is released under the Apache 2.0 license.
+
+A copy of Apache 2.0 is included in this file.
+
+StyleGAN2
+The codes are modified from the repository stylegan2-pytorch. Many thanks to the author - Kim Seonghyeon 😊 for translating from the official TensorFlow codes to PyTorch ones. Here is the license of stylegan2-pytorch.
+The official repository is https://github.com/NVlabs/stylegan2, and here is the NVIDIA license.
+DFDNet
+The codes are largely modified from the repository DFDNet. Their license is Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
+
+Terms of the Nvidia License:
+---------------------------------------------
+
+1. Definitions
+
+"Licensor" means any person or entity that distributes its Work.
+
+"Software" means the original work of authorship made available under
+this License.
+
+"Work" means the Software and any additions to or derivative works of
+the Software that are made available under this License.
+
+"Nvidia Processors" means any central processing unit (CPU), graphics
+processing unit (GPU), field-programmable gate array (FPGA),
+application-specific integrated circuit (ASIC) or any combination
+thereof designed, made, sold, or provided by Nvidia or its affiliates.
+
+The terms "reproduce," "reproduction," "derivative works," and
+"distribution" have the meaning as provided under U.S. copyright law;
+provided, however, that for the purposes of this License, derivative
+works shall not include works that remain separable from, or merely
+link (or bind by name) to the interfaces of, the Work.
+
+Works, including the Software, are "made available" under this License
+by including in or with the Work either (a) a copyright notice
+referencing the applicability of this License to the Work, or (b) a
+copy of this License.
+
+2. License Grants
+
+    2.1 Copyright Grant. Subject to the terms and conditions of this
+    License, each Licensor grants to you a perpetual, worldwide,
+    non-exclusive, royalty-free, copyright license to reproduce,
+    prepare derivative works of, publicly display, publicly perform,
+    sublicense and distribute its Work and any resulting derivative
+    works in any form.
+
+3. Limitations
+
+    3.1 Redistribution. You may reproduce or distribute the Work only
+    if (a) you do so under this License, (b) you include a complete
+    copy of this License with your distribution, and (c) you retain
+    without modification any copyright, patent, trademark, or
+    attribution notices that are present in the Work.
+
+    3.2 Derivative Works. You may specify that additional or different
+    terms apply to the use, reproduction, and distribution of your
+    derivative works of the Work ("Your Terms") only if (a) Your Terms
+    provide that the use limitation in Section 3.3 applies to your
+    derivative works, and (b) you identify the specific derivative
+    works that are subject to Your Terms. Notwithstanding Your Terms,
+    this License (including the redistribution requirements in Section
+    3.1) will continue to apply to the Work itself.
+
+    3.3 Use Limitation. The Work and any derivative works thereof only
+    may be used or intended for use non-commercially. The Work or
+    derivative works thereof may be used or intended for use by Nvidia
+    or its affiliates commercially or non-commercially. As used herein,
+    "non-commercially" means for research or evaluation purposes only.
+
+    3.4 Patent Claims. If you bring or threaten to bring a patent claim
+    against any Licensor (including any claim, cross-claim or
+    counterclaim in a lawsuit) to enforce any patents that you allege
+    are infringed by any Work, then your rights under this License from
+    such Licensor (including the grants in Sections 2.1 and 2.2) will
+    terminate immediately.
+
+    3.5 Trademarks. This License does not grant any rights to use any
+    Licensor's or its affiliates' names, logos, or trademarks, except
+    as necessary to reproduce the notices described in this License.
+
+    3.6 Termination. If you violate any term of this License, then your
+    rights under this License (including the grants in Sections 2.1 and
+    2.2) will terminate immediately.
+
+4. Disclaimer of Warranty.
+
+THE WORK IS PROVIDED "AS IS" WITHOUT WARRANTIES OR CONDITIONS OF ANY
+KIND, EITHER EXPRESS OR IMPLIED, INCLUDING WARRANTIES OR CONDITIONS OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, TITLE OR
+NON-INFRINGEMENT. YOU BEAR THE RISK OF UNDERTAKING ANY ACTIVITIES UNDER
+THIS LICENSE.
+
+5. Limitation of Liability.
+
+EXCEPT AS PROHIBITED BY APPLICABLE LAW, IN NO EVENT AND UNDER NO LEGAL
+THEORY, WHETHER IN TORT (INCLUDING NEGLIGENCE), CONTRACT, OR OTHERWISE
+SHALL ANY LICENSOR BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY DIRECT,
+INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING OUT OF
+OR RELATED TO THIS LICENSE, THE USE OR INABILITY TO USE THE WORK
+(INCLUDING BUT NOT LIMITED TO LOSS OF GOODWILL, BUSINESS INTERRUPTION,
+LOST PROFITS OR DATA, COMPUTER FAILURE OR MALFUNCTION, OR ANY OTHER
+COMMERCIAL DAMAGES OR LOSSES), EVEN IF THE LICENSOR HAS BEEN ADVISED OF
+THE POSSIBILITY OF SUCH DAMAGES.
+
+MIT License
+
+Copyright (c) 2019 Kim Seonghyeon
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+
+
+
+Open Source Software licensed under the BSD 3-Clause license:
+---------------------------------------------
+1. torchvision
+Copyright (c) Soumith Chintala 2016,
+All rights reserved.
+
+2. torch
+Copyright (c) 2016-     Facebook, Inc            (Adam Paszke)
+Copyright (c) 2014-     Facebook, Inc            (Soumith Chintala)
+Copyright (c) 2011-2014 Idiap Research Institute (Ronan Collobert)
+Copyright (c) 2012-2014 Deepmind Technologies    (Koray Kavukcuoglu)
+Copyright (c) 2011-2012 NEC Laboratories America (Koray Kavukcuoglu)
+Copyright (c) 2011-2013 NYU                      (Clement Farabet)
+Copyright (c) 2006-2010 NEC Laboratories America (Ronan Collobert, Leon Bottou, Iain Melvin, Jason Weston)
+Copyright (c) 2006      Idiap Research Institute (Samy Bengio)
+Copyright (c) 2001-2004 Idiap Research Institute (Ronan Collobert, Samy Bengio, Johnny Mariethoz)
+
+
+Terms of the BSD 3-Clause License:
+---------------------------------------------
+Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
+
+1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
+
+3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+
+
+Open Source Software licensed under the BSD 3-Clause License and Other Licenses of the Third-Party Components therein:
+---------------------------------------------
+1. numpy
+Copyright (c) 2005-2020, NumPy Developers.
+All rights reserved.
+
+A copy of BSD 3-Clause License is included in this file.
+
+The NumPy repository and source distributions bundle several libraries that are
+compatibly licensed.  We list these here.
+
+Name: Numpydoc
+Files: doc/sphinxext/numpydoc/*
+License: BSD-2-Clause
+  For details, see doc/sphinxext/LICENSE.txt
+
+Name: scipy-sphinx-theme
+Files: doc/scipy-sphinx-theme/*
+License: BSD-3-Clause AND PSF-2.0 AND Apache-2.0
+  For details, see doc/scipy-sphinx-theme/LICENSE.txt
+
+Name: lapack-lite
+Files: numpy/linalg/lapack_lite/*
+License: BSD-3-Clause
+  For details, see numpy/linalg/lapack_lite/LICENSE.txt
+
+Name: tempita
+Files: tools/npy_tempita/*
+License: MIT
+  For details, see tools/npy_tempita/license.txt
+
+Name: dragon4
+Files: numpy/core/src/multiarray/dragon4.c
+License: MIT
+  For license text, see numpy/core/src/multiarray/dragon4.c
+
+
+
+Open Source Software licensed under the MIT license:
+---------------------------------------------
+1. facexlib
+Copyright (c) 2020 Xintao Wang
+
+2. opencv-python
+Copyright (c) Olli-Pekka Heinisuo
+Please note that only files in cv2 package are used.
+
+
+Terms of the MIT License:
+---------------------------------------------
+Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+
+
+Open Source Software licensed under the MIT license and Other Licenses of the Third-Party Components therein:
+---------------------------------------------
+1. tqdm
+Copyright (c) 2013 noamraph
+
+`tqdm` is a product of collaborative work.
+Unless otherwise stated, all authors (see commit logs) retain copyright
+for their respective work, and release the work under the MIT licence
+(text below).
+
+Exceptions or notable authors are listed below
+in reverse chronological order:
+
+* files: *
+  MPLv2.0 2015-2020 (c) Casper da Costa-Luis
+  [casperdcl](https://github.com/casperdcl).
+* files: tqdm/_tqdm.py
+  MIT 2016 (c) [PR #96] on behalf of Google Inc.
+* files: tqdm/_tqdm.py setup.py README.rst MANIFEST.in .gitignore
+  MIT 2013 (c) Noam Yorav-Raphael, original author.
+
+[PR #96]: https://github.com/tqdm/tqdm/pull/96
+
+
+Mozilla Public Licence (MPL) v. 2.0 - Exhibit A
+-----------------------------------------------
+
+This Source Code Form is subject to the terms of the
+Mozilla Public License, v. 2.0.
+If a copy of the MPL was not distributed with this file,
+You can obtain one at https://mozilla.org/MPL/2.0/.
+
+
+MIT License (MIT)
+-----------------
+
+Copyright (c) 2013 noamraph
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of
+this software and associated documentation files (the "Software"), to deal in
+the Software without restriction, including without limitation the rights to
+use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of
+the Software, and to permit persons to whom the Software is furnished to do so,
+subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
+FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
+COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
+IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

backend/headless/fcbh_extras/chainner_models/architecture/face/LICENSE-codeformer

@@ -0,0 +1,35 @@
+S-Lab License 1.0
+
+Copyright 2022 S-Lab
+
+Redistribution and use for non-commercial purpose in source and
+binary forms, with or without modification, are permitted provided
+that the following conditions are met:
+
+1. Redistributions of source code must retain the above copyright
+   notice, this list of conditions and the following disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright
+   notice, this list of conditions and the following disclaimer in
+   the documentation and/or other materials provided with the
+   distribution.
+
+3. Neither the name of the copyright holder nor the names of its
+   contributors may be used to endorse or promote products derived
+   from this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+In the event that redistribution and/or use for commercial purpose in
+source or binary forms, with or without modification is required,
+please contact the contributor(s) of the work.

backend/headless/fcbh_extras/chainner_models/architecture/face/arcface_arch.py

@@ -0,0 +1,265 @@
+import torch.nn as nn
+
+
+def conv3x3(inplanes, outplanes, stride=1):
+    """A simple wrapper for 3x3 convolution with padding.
+
+    Args:
+        inplanes (int): Channel number of inputs.
+        outplanes (int): Channel number of outputs.
+        stride (int): Stride in convolution. Default: 1.
+    """
+    return nn.Conv2d(
+        inplanes, outplanes, kernel_size=3, stride=stride, padding=1, bias=False
+    )
+
+
+class BasicBlock(nn.Module):
+    """Basic residual block used in the ResNetArcFace architecture.
+
+    Args:
+        inplanes (int): Channel number of inputs.
+        planes (int): Channel number of outputs.
+        stride (int): Stride in convolution. Default: 1.
+        downsample (nn.Module): The downsample module. Default: None.
+    """
+
+    expansion = 1  # output channel expansion ratio
+
+    def __init__(self, inplanes, planes, stride=1, downsample=None):
+        super(BasicBlock, self).__init__()
+        self.conv1 = conv3x3(inplanes, planes, stride)
+        self.bn1 = nn.BatchNorm2d(planes)
+        self.relu = nn.ReLU(inplace=True)
+        self.conv2 = conv3x3(planes, planes)
+        self.bn2 = nn.BatchNorm2d(planes)
+        self.downsample = downsample
+        self.stride = stride
+
+    def forward(self, x):
+        residual = x
+
+        out = self.conv1(x)
+        out = self.bn1(out)
+        out = self.relu(out)
+
+        out = self.conv2(out)
+        out = self.bn2(out)
+
+        if self.downsample is not None:
+            residual = self.downsample(x)
+
+        out += residual
+        out = self.relu(out)
+
+        return out
+
+
+class IRBlock(nn.Module):
+    """Improved residual block (IR Block) used in the ResNetArcFace architecture.
+
+    Args:
+        inplanes (int): Channel number of inputs.
+        planes (int): Channel number of outputs.
+        stride (int): Stride in convolution. Default: 1.
+        downsample (nn.Module): The downsample module. Default: None.
+        use_se (bool): Whether use the SEBlock (squeeze and excitation block). Default: True.
+    """
+
+    expansion = 1  # output channel expansion ratio
+
+    def __init__(self, inplanes, planes, stride=1, downsample=None, use_se=True):
+        super(IRBlock, self).__init__()
+        self.bn0 = nn.BatchNorm2d(inplanes)
+        self.conv1 = conv3x3(inplanes, inplanes)
+        self.bn1 = nn.BatchNorm2d(inplanes)
+        self.prelu = nn.PReLU()
+        self.conv2 = conv3x3(inplanes, planes, stride)
+        self.bn2 = nn.BatchNorm2d(planes)
+        self.downsample = downsample
+        self.stride = stride
+        self.use_se = use_se
+        if self.use_se:
+            self.se = SEBlock(planes)
+
+    def forward(self, x):
+        residual = x
+        out = self.bn0(x)
+        out = self.conv1(out)
+        out = self.bn1(out)
+        out = self.prelu(out)
+
+        out = self.conv2(out)
+        out = self.bn2(out)
+        if self.use_se:
+            out = self.se(out)
+
+        if self.downsample is not None:
+            residual = self.downsample(x)
+
+        out += residual
+        out = self.prelu(out)
+
+        return out
+
+
+class Bottleneck(nn.Module):
+    """Bottleneck block used in the ResNetArcFace architecture.
+
+    Args:
+        inplanes (int): Channel number of inputs.
+        planes (int): Channel number of outputs.
+        stride (int): Stride in convolution. Default: 1.
+        downsample (nn.Module): The downsample module. Default: None.
+    """
+
+    expansion = 4  # output channel expansion ratio
+
+    def __init__(self, inplanes, planes, stride=1, downsample=None):
+        super(Bottleneck, self).__init__()
+        self.conv1 = nn.Conv2d(inplanes, planes, kernel_size=1, bias=False)
+        self.bn1 = nn.BatchNorm2d(planes)
+        self.conv2 = nn.Conv2d(
+            planes, planes, kernel_size=3, stride=stride, padding=1, bias=False
+        )
+        self.bn2 = nn.BatchNorm2d(planes)
+        self.conv3 = nn.Conv2d(
+            planes, planes * self.expansion, kernel_size=1, bias=False
+        )
+        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
+        self.relu = nn.ReLU(inplace=True)
+        self.downsample = downsample
+        self.stride = stride
+
+    def forward(self, x):
+        residual = x
+
+        out = self.conv1(x)
+        out = self.bn1(out)
+        out = self.relu(out)
+
+        out = self.conv2(out)
+        out = self.bn2(out)
+        out = self.relu(out)
+
+        out = self.conv3(out)
+        out = self.bn3(out)
+
+        if self.downsample is not None:
+            residual = self.downsample(x)
+
+        out += residual
+        out = self.relu(out)
+
+        return out
+
+
+class SEBlock(nn.Module):
+    """The squeeze-and-excitation block (SEBlock) used in the IRBlock.
+
+    Args:
+        channel (int): Channel number of inputs.
+        reduction (int): Channel reduction ration. Default: 16.
+    """
+
+    def __init__(self, channel, reduction=16):
+        super(SEBlock, self).__init__()
+        self.avg_pool = nn.AdaptiveAvgPool2d(
+            1
+        )  # pool to 1x1 without spatial information
+        self.fc = nn.Sequential(
+            nn.Linear(channel, channel // reduction),
+            nn.PReLU(),
+            nn.Linear(channel // reduction, channel),
+            nn.Sigmoid(),
+        )
+
+    def forward(self, x):
+        b, c, _, _ = x.size()
+        y = self.avg_pool(x).view(b, c)
+        y = self.fc(y).view(b, c, 1, 1)
+        return x * y
+
+
+class ResNetArcFace(nn.Module):
+    """ArcFace with ResNet architectures.
+
+    Ref: ArcFace: Additive Angular Margin Loss for Deep Face Recognition.
+
+    Args:
+        block (str): Block used in the ArcFace architecture.
+        layers (tuple(int)): Block numbers in each layer.
+        use_se (bool): Whether use the SEBlock (squeeze and excitation block). Default: True.
+    """
+
+    def __init__(self, block, layers, use_se=True):
+        if block == "IRBlock":
+            block = IRBlock
+        self.inplanes = 64
+        self.use_se = use_se
+        super(ResNetArcFace, self).__init__()
+
+        self.conv1 = nn.Conv2d(1, 64, kernel_size=3, padding=1, bias=False)
+        self.bn1 = nn.BatchNorm2d(64)
+        self.prelu = nn.PReLU()
+        self.maxpool = nn.MaxPool2d(kernel_size=2, stride=2)
+        self.layer1 = self._make_layer(block, 64, layers[0])
+        self.layer2 = self._make_layer(block, 128, layers[1], stride=2)
+        self.layer3 = self._make_layer(block, 256, layers[2], stride=2)
+        self.layer4 = self._make_layer(block, 512, layers[3], stride=2)
+        self.bn4 = nn.BatchNorm2d(512)
+        self.dropout = nn.Dropout()
+        self.fc5 = nn.Linear(512 * 8 * 8, 512)
+        self.bn5 = nn.BatchNorm1d(512)
+
+        # initialization
+        for m in self.modules():
+            if isinstance(m, nn.Conv2d):
+                nn.init.xavier_normal_(m.weight)
+            elif isinstance(m, nn.BatchNorm2d) or isinstance(m, nn.BatchNorm1d):
+                nn.init.constant_(m.weight, 1)
+                nn.init.constant_(m.bias, 0)
+            elif isinstance(m, nn.Linear):
+                nn.init.xavier_normal_(m.weight)
+                nn.init.constant_(m.bias, 0)
+
+    def _make_layer(self, block, planes, num_blocks, stride=1):
+        downsample = None
+        if stride != 1 or self.inplanes != planes * block.expansion:
+            downsample = nn.Sequential(
+                nn.Conv2d(
+                    self.inplanes,
+                    planes * block.expansion,
+                    kernel_size=1,
+                    stride=stride,
+                    bias=False,
+                ),
+                nn.BatchNorm2d(planes * block.expansion),
+            )
+        layers = []
+        layers.append(
+            block(self.inplanes, planes, stride, downsample, use_se=self.use_se)
+        )
+        self.inplanes = planes
+        for _ in range(1, num_blocks):
+            layers.append(block(self.inplanes, planes, use_se=self.use_se))
+
+        return nn.Sequential(*layers)
+
+    def forward(self, x):
+        x = self.conv1(x)
+        x = self.bn1(x)
+        x = self.prelu(x)
+        x = self.maxpool(x)
+
+        x = self.layer1(x)
+        x = self.layer2(x)
+        x = self.layer3(x)
+        x = self.layer4(x)
+        x = self.bn4(x)
+        x = self.dropout(x)
+        x = x.view(x.size(0), -1)
+        x = self.fc5(x)
+        x = self.bn5(x)
+
+        return x