Release verified Image21-INT4 conversion

.gitattributes

@@ -33,3 +33,33 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/chinese_text-s123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/chinese_text-s42.png filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/composition-s123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/composition-s42.png filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/edit-s1000042.png filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/edit-s1000123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/english_text-s123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/english_text-s42.png filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/portrait-s123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/portrait-s42.png filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/rgba-s123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/rgba-s42.png filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/texture-s123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/bf16/texture-s42.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/chinese_text-s123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/chinese_text-s42.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/composition-s123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/composition-s42.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/edit-s1000042.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/edit-s1000123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/english_text-s123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/english_text-s42.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/portrait-s123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/portrait-s42.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/rgba-s123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/rgba-s42.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/texture-s123.png filter=lfs diff=lfs merge=lfs -text
+evaluation/int4/texture-s42.png filter=lfs diff=lfs merge=lfs -text
+evaluation/vram8/portrait-s42.png filter=lfs diff=lfs merge=lfs -text
+processor/tokenizer.json filter=lfs diff=lfs merge=lfs -text

CHANGES.md

@@ -0,0 +1,9 @@
+# Modifications
+
+Modified by ixim / iximbox: eligible linear weights converted from Qwen-Image-2.1 to SDNQ UINT4 with SVD rank 32. Built with Qwen. Non-commercial research/evaluation under the accompanying Qwen Research License.
+
+- Converted eligible transformer and text-encoder linear layers to SDNQ UINT4.
+- Stored a rank-32 SVD residual of the quantization error with the weights.
+- Left the requested sensitive projections, normalization, embeddings, vision tower, output head and VAE in floating point.
+- Did not use a calibration set or fine-tuning.
+- Quantized matmul is off so CUDA and Apple Silicon use the same eager dequantization.

LICENSE

@@ -0,0 +1,55 @@
+Qwen RESEARCH LICENSE AGREEMENT
+
+Qwen RESEARCH LICENSE AGREEMENT Release Date: September 20, 2026
+
+By clicking to agree or by using or distributing any portion or element of the Qwen Materials, you will be deemed to have recognized and accepted the content of this Agreement, which is effective immediately.
+
+1. Definitions
+    a. This Qwen RESEARCH LICENSE AGREEMENT (this "Agreement") shall mean the terms and conditions for use, reproduction, distribution and modification of the Materials as defined by this Agreement.
+    b. "We" (or "Us") shall mean Hangzhou Tongyi Laboratory Technology Co., Ltd.
+    c. "You" (or "Your") shall mean a natural person or legal entity exercising the rights granted by this Agreement and/or using the Materials for any purpose and in any field of use.
+    d. "Third Parties" shall mean individuals or legal entities that are not under common control with us or you.
+    e. "Qwen" shall mean the large language models, diffusion models, and software and algorithms, consisting of trained model weights, parameters (including optimizer states), machine-learning model code, inference-enabling code, training-enabling code, fine-tuning enabling code and other elements of the foregoing distributed by us.
+    f. "Materials" shall mean, collectively, our proprietary Qwen and Documentation (and any portion thereof) made available under this Agreement.
+    g. "Source" form shall mean the preferred form for making modifications, including but not limited to model source code, documentation source, and configuration files.
+    h. "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.
+    i. "Non-Commercial" shall mean for research or evaluation purposes only.
+
+2. Grant of Rights
+    a. You are granted a non-exclusive, worldwide, non-transferable and royalty-free limited license under our intellectual property or other rights owned by us embodied in the Materials to use, reproduce, distribute, copy, create derivative works of, and make modifications to the Materials FOR NON-COMMERCIAL PURPOSES ONLY. 
+    b. You shall not use the Materials for any commercial purpose without obtaining a separate commercial license from us. If you wish to use the Materials commercially, you shall request a license from us at model-business@notice.qwencloud.com.
+
+3. Redistribution
+Subject to Section 2 (Grant of Rights), you may distribute copies or make the Materials, or derivative works thereof, available as part of a product or service that contains any of them, with or without modifications, and in Source or Object form, provided that you meet the following conditions:
+    a. You shall give any other recipients of the Materials or derivative works a copy of this Agreement;
+    b. You shall cause any modified files to carry prominent notices stating that you changed the files;
+    c. You shall retain in all copies of the Materials that you distribute the following attribution notices within a "Notice" text file distributed as a part of such copies: "Qwen is licensed under the Qwen RESEARCH LICENSE AGREEMENT, Copyright (c) 2026 Hangzhou Tongyi Laboratory Technology Co., Ltd. All Rights Reserved."; and
+    d. 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 terms and conditions of this Agreement.
+
+4. Rules of use
+    a. The Materials may be subject to export controls or restrictions in China, the United States or other countries or regions. You shall comply with applicable laws and regulations in your use of the Materials.
+    b. If you use the Materials or any outputs or results therefrom to create, train, fine-tune, or improve an AI model that is distributed or made available, you shall prominently display “Built with Qwen” or “Improved using Qwen” in the related product documentation.
+    c. You shall not use "Qwen" as the primary name or identifier of any derivative works or products; reasonable descriptive use (e.g., "fine-tuned from Qwen Image") is permitted.
+
+5. Intellectual Property
+    a. We retain ownership of all intellectual property rights in and to the Materials and derivatives made by or for us. Conditioned upon compliance with the terms and conditions of this Agreement, with respect to any derivative works and modifications of the Materials that are made by you, you are and will be the owner of such derivative works and modifications.
+    b. No trademark license is granted to use the trade names, trademarks, service marks, or product names of us, except as required to fulfill notice requirements under this Agreement or as required for reasonable and customary use in describing and redistributing the Materials.
+    c. If you commence a lawsuit or other proceedings (including a cross-claim or counterclaim in a lawsuit) against us or any entity alleging that the Materials or any output therefrom, or any part of the foregoing, infringe any intellectual property or other right owned or licensable by you, then all licenses granted to you under this Agreement shall terminate as of the date such lawsuit or other proceeding is commenced or brought.
+
+6. Disclaimer of Warranty and Limitation of Liability
+    a. We are not obligated to support, update, provide training for, or develop any further version of the Qwen Materials or to grant any license thereto.
+    b. THE MATERIALS ARE PROVIDED "AS IS" WITHOUT ANY EXPRESS OR IMPLIED WARRANTY OF ANY KIND INCLUDING WARRANTIES OF MERCHANTABILITY, NONINFRINGEMENT, OR FITNESS FOR A PARTICULAR PURPOSE. WE MAKE NO WARRANTY AND ASSUME NO RESPONSIBILITY FOR THE SAFETY OR STABILITY OF THE MATERIALS AND ANY OUTPUT THEREFROM.
+    c. IN NO EVENT SHALL WE BE LIABLE TO YOU FOR ANY DAMAGES, INCLUDING, BUT NOT LIMITED TO ANY DIRECT, OR INDIRECT, SPECIAL OR CONSEQUENTIAL DAMAGES ARISING FROM YOUR USE OR INABILITY TO USE THE MATERIALS OR ANY OUTPUT OF IT, NO MATTER HOW IT’S CAUSED.
+    d. You will defend, indemnify and hold harmless us from and against any claim by any third party arising out of or related to your use or distribution of the Materials.
+
+7. Survival and Termination.
+    a. The term of this Agreement shall commence upon your acceptance of this Agreement or access to the Materials and will continue in full force and effect until terminated in accordance with the terms and conditions herein.
+    b. We may terminate this Agreement if you breach any of the terms or conditions of this Agreement. Upon termination of this Agreement, you must delete and cease use of the Materials. Sections 6 and 8 shall survive the termination of this Agreement.
+
+8. Governing Law and Jurisdiction.
+    a. This Agreement and any dispute arising out of or relating to it will be governed by the laws of China, without regard to conflict of law principles, and the UN Convention on Contracts for the International Sale of Goods does not apply to this Agreement.
+    b. The People's Courts in Hangzhou City shall have exclusive jurisdiction over any dispute arising out of this Agreement.
+
+9. Other Terms and Conditions.
+    a. Any arrangements, understandings, or agreements regarding the Material not stated herein are separate from and independent of the terms and conditions of this Agreement. You shall request a separate license from us, if you use the Materials in ways not expressly agreed to in this Agreement. 
+    b. We shall not be bound by any additional or different terms or conditions communicated by you unless expressly agreed.

MANIFEST.json

@@ -0,0 +1,462 @@
+[
+  {
+    "path": "benchmarks/cases.json",
+    "size": 1962,
+    "sha256": "c0bc77b0aa28c9f11442c30cafd07d31bca72a179c7b759fb2a5ed9e328896c8"
+  },
+  {
+    "path": "cards/huggingface.md",
+    "size": 3674,
+    "sha256": "ad9d4f47164fbe91adeab1596fd1643e011ddb70b30b337695a3c37b9adf3479"
+  },
+  {
+    "path": "cards/modelscope.md",
+    "size": 3498,
+    "sha256": "f837db6386c4d35b68505ca16bc3d8bb383b7767214d7de59f9c88888d802602"
+  },
+  {
+    "path": "CHANGES.md",
+    "size": 653,
+    "sha256": "ef6272bb35b8896bda8e4b359a2c593ae20516ef8925707b79f4ef7ce1094501"
+  },
+  {
+    "path": "conversion.json",
+    "size": 1079,
+    "sha256": "7b50874c8e58b499db338d8d13aa83f62f1c817136d5425cc0611fe1cd8d6fe3"
+  },
+  {
+    "path": "evaluation/bf16/chinese_text-s123.png",
+    "size": 1194079,
+    "sha256": "eae74256d7ed1b3519d39c7d58961cf7a937224d19a3c3239ca7eba4178edf49"
+  },
+  {
+    "path": "evaluation/bf16/chinese_text-s42.png",
+    "size": 1303893,
+    "sha256": "7082cb71d99705bc42e27e2c206b13e3beb9e62cfd6304cfabe203b82641692c"
+  },
+  {
+    "path": "evaluation/bf16/composition-s123.png",
+    "size": 1220815,
+    "sha256": "bc11612f260acd8af0d834f3b28c752cefd0437d24c9a01dae0d46647bdf29ae"
+  },
+  {
+    "path": "evaluation/bf16/composition-s42.png",
+    "size": 1194343,
+    "sha256": "91e967685fbbd79ba13daece22a0f72c863cb518db88d5d2a6f9aecfc5f292b1"
+  },
+  {
+    "path": "evaluation/bf16/edit-s1000042.png",
+    "size": 1813053,
+    "sha256": "a846f593fd08c74231fa6154654dc089668c2e4f2359b6e65f5d136cbe8fc493"
+  },
+  {
+    "path": "evaluation/bf16/edit-s1000123.png",
+    "size": 1818914,
+    "sha256": "e46d6f1316e09816be7d980a65277e53fbb5ba25578e2f6dfcd7e890d1b14b32"
+  },
+  {
+    "path": "evaluation/bf16/english_text-s123.png",
+    "size": 905011,
+    "sha256": "cf5e76dee9f0e403ebdf24a64e1fcd604ee2b6c8e645a032fba2807acc34adf2"
+  },
+  {
+    "path": "evaluation/bf16/english_text-s42.png",
+    "size": 969130,
+    "sha256": "ba31b0c8a312e96408096566013c667b3a32bd7720749ca47120960b17b41ca6"
+  },
+  {
+    "path": "evaluation/bf16/environment.json",
+    "size": 8815,
+    "sha256": "f2f4ca7f466cada65f482d4d08f106a74545c0d18269123eb355fe25b205412f"
+  },
+  {
+    "path": "evaluation/bf16/portrait-s123.png",
+    "size": 1930022,
+    "sha256": "cff376e50c41c4aee2b9e4aabc502f4791774f6ad285d8a7673a61ebb1fa650d"
+  },
+  {
+    "path": "evaluation/bf16/portrait-s42.png",
+    "size": 1868618,
+    "sha256": "01e37e178fb0d35d5c706b7edd76bc1a6c5b50ec3536901a467fd45295d097aa"
+  },
+  {
+    "path": "evaluation/bf16/records.jsonl",
+    "size": 14333,
+    "sha256": "17d1c4f42e3a0c22105f7e86079642630ea8953608e69aad7bd538cb2e170c02"
+  },
+  {
+    "path": "evaluation/bf16/rgba-s123.png",
+    "size": 926957,
+    "sha256": "6f1abbb1a64ac553fc576a0250d370ab5bb7e6d5e6372f42f67045a5bb0951a6"
+  },
+  {
+    "path": "evaluation/bf16/rgba-s42.png",
+    "size": 850131,
+    "sha256": "6f5dfea6ba02e41cbf20617b6e863e1adfa139d79fad2ca5c64dfc815f591d82"
+  },
+  {
+    "path": "evaluation/bf16/texture-s123.png",
+    "size": 1760350,
+    "sha256": "18c4403f388d0e03447473c03497141169f333fddeaeb40227163dc3b9d68d3e"
+  },
+  {
+    "path": "evaluation/bf16/texture-s42.png",
+    "size": 1797822,
+    "sha256": "40a0f73bb407a394fa9e8a3f6922455df716b9725ad07437fa10c1daca2c2095"
+  },
+  {
+    "path": "evaluation/comparison.csv",
+    "size": 2206,
+    "sha256": "c7bf94da06b673b1fce8fdd82dc903a9e8c42c6609c5a81b313355888b622447"
+  },
+  {
+    "path": "evaluation/comparison.html",
+    "size": 6805,
+    "sha256": "88a72e1c8d9465d5d4497271249e15bd927926160a74ff1420ca3de721610033"
+  },
+  {
+    "path": "evaluation/int4/chinese_text-s123.png",
+    "size": 1252178,
+    "sha256": "b1740259fff3ebc07d14c12a837a899267a468ff4318a53c163815a3d6be53fb"
+  },
+  {
+    "path": "evaluation/int4/chinese_text-s42.png",
+    "size": 1319559,
+    "sha256": "61387812385423c8c240d7f34875fd3b2b88cc1662db3ce084757c434915d1e8"
+  },
+  {
+    "path": "evaluation/int4/composition-s123.png",
+    "size": 1120600,
+    "sha256": "c7b45122d69e8b4720ab9e3877f86a8da172629ccc0335ce303195b5000944e8"
+  },
+  {
+    "path": "evaluation/int4/composition-s42.png",
+    "size": 1089490,
+    "sha256": "2074e38fd2af7d7cc8cf2f483553fdfbeeaad65094e70062462c0fa9b48b6f29"
+  },
+  {
+    "path": "evaluation/int4/edit-s1000042.png",
+    "size": 1829190,
+    "sha256": "ae986b19def44d5bb3fa42e0338322a5e15177dfd7b909c4241c0c299f7c0f7e"
+  },
+  {
+    "path": "evaluation/int4/edit-s1000123.png",
+    "size": 1840221,
+    "sha256": "c60c71631e6e0522d4ef29b13be4354231a09a14d3768881e9cfee328c6dc491"
+  },
+  {
+    "path": "evaluation/int4/english_text-s123.png",
+    "size": 892793,
+    "sha256": "add2118fcebc3e2f63788382b323813fe7d967f1e88c91ae271fec057156e3a1"
+  },
+  {
+    "path": "evaluation/int4/english_text-s42.png",
+    "size": 976444,
+    "sha256": "d6880ef09d3d5986da086e012b89e4a5fabfde16e16721e49c0436baadec53da"
+  },
+  {
+    "path": "evaluation/int4/environment.json",
+    "size": 7981,
+    "sha256": "a2b6db3e4e868212f09cb2a4759d8df385e20fec7767a0cb6a35b77d29009c0a"
+  },
+  {
+    "path": "evaluation/int4/portrait-s123.png",
+    "size": 1966813,
+    "sha256": "0f3b5e3c9c2464d3f2567c8f6e9040494de1fdf06042fcb0696df58b8f58e047"
+  },
+  {
+    "path": "evaluation/int4/portrait-s42.png",
+    "size": 1910863,
+    "sha256": "a084dccd83c651b709a301925fcc085f005d2b4f49fac3ff0817d6d78644e7d3"
+  },
+  {
+    "path": "evaluation/int4/records.jsonl",
+    "size": 14312,
+    "sha256": "c81e87e75d15575d12b57d1c63a865f40e0023da66ea9ae55c077922551218bc"
+  },
+  {
+    "path": "evaluation/int4/rgba-s123.png",
+    "size": 959844,
+    "sha256": "fb13ec3ed4d00651e1050c4b1c0f08ef47cd85f179e3b3bf516d58dde6dcd2ef"
+  },
+  {
+    "path": "evaluation/int4/rgba-s42.png",
+    "size": 858470,
+    "sha256": "47635b9cf96c9d47991df7c39b2bf8a5bd7c4b8e05ddea33034b83eff7e10f7c"
+  },
+  {
+    "path": "evaluation/int4/texture-s123.png",
+    "size": 1772861,
+    "sha256": "7c50f70eb4f31c8f4322330b40a2cfd892640157abcc2f4ded826b6f7b8345f6"
+  },
+  {
+    "path": "evaluation/int4/texture-s42.png",
+    "size": 1749173,
+    "sha256": "2f4809ffb0301e2ddddf66ddfb834d1e750175a675fb05067cb671292ff50187"
+  },
+  {
+    "path": "evaluation/qualitative.md",
+    "size": 3941,
+    "sha256": "3b90f2ebcc3ccd85bf9d03266c01c0d7bd5a1f731072a4dad58aa4860c799ea5"
+  },
+  {
+    "path": "evaluation/report.md",
+    "size": 1622,
+    "sha256": "e802bdac15dc1afeebfbad8c0de2930b525cf24d0d34b906119985a53ba1be50"
+  },
+  {
+    "path": "evaluation/summary.json",
+    "size": 364,
+    "sha256": "d91fbc1cc96b6df2e6ef9d5568e410ce53b4920a174e09c5a026c09de7a7b735"
+  },
+  {
+    "path": "evaluation/vram8/environment.json",
+    "size": 8068,
+    "sha256": "1f4720fce9a987b75218509f2a8a99d893fdd7a08d1e29f40a2ffed57bbeca0b"
+  },
+  {
+    "path": "evaluation/vram8/portrait-s42.png",
+    "size": 1895579,
+    "sha256": "b1b00f13a0671ce7d751fd964dc6735225eda168f30ce3c8eb99d80786f5fce3"
+  },
+  {
+    "path": "evaluation/vram8/records.jsonl",
+    "size": 1024,
+    "sha256": "b05e777da92280e188d3adcb25c3dcc0224d29dd8457b313721532eea0d4ffb1"
+  },
+  {
+    "path": "LICENSE",
+    "size": 7831,
+    "sha256": "8dc973f024ff95966bea25866efa443fd16776dcb1001e681e3d467ea572b28d"
+  },
+  {
+    "path": "model_index.json",
+    "size": 447,
+    "sha256": "cf1ecd104ea090855d60d8cec0895c1e9d6ee41b2f87e231b678beecd1cf7809"
+  },
+  {
+    "path": "Notice",
+    "size": 540,
+    "sha256": "5f1439a05fbaa3082014c8a84b615754eae3897a40ac40eecfeba8857b3f9412"
+  },
+  {
+    "path": "processor/added_tokens.json",
+    "size": 707,
+    "sha256": "c0284b582e14987fbd3d5a2cb2bd139084371ed9acbae488829a1c900833c680"
+  },
+  {
+    "path": "processor/chat_template.jinja",
+    "size": 5292,
+    "sha256": "3636d0f0bd6bef02654cdffdc447b79cb2cef8ab02cc75267345946291a489e4"
+  },
+  {
+    "path": "processor/merges.txt",
+    "size": 1671853,
+    "sha256": "8831e4f1a044471340f7c0a83d7bd71306a5b867e95fd870f74d0c5308a904d5"
+  },
+  {
+    "path": "processor/preprocessor_config.json",
+    "size": 782,
+    "sha256": "93585062a80db5e8ca038efc7726a3e6411d9db948472d81d63c6303993be8c5"
+  },
+  {
+    "path": "processor/special_tokens_map.json",
+    "size": 613,
+    "sha256": "76862e765266b85aa9459767e33cbaf13970f327a0e88d1c65846c2ddd3a1ecd"
+  },
+  {
+    "path": "processor/tokenizer.json",
+    "size": 11422654,
+    "sha256": "aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4"
+  },
+  {
+    "path": "processor/tokenizer_config.json",
+    "size": 5445,
+    "sha256": "81ec7bb9530159b326c0bef1d0b6c33d392090524014ea3f0123a3c1eb9c2af5"
+  },
+  {
+    "path": "processor/video_preprocessor_config.json",
+    "size": 817,
+    "sha256": "59c5c9eb52182eb14c06ffb10ca9effd29adce5f238a95de23ca14a38dbd2cb1"
+  },
+  {
+    "path": "processor/vocab.json",
+    "size": 2776833,
+    "sha256": "ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910"
+  },
+  {
+    "path": "PUBLISHING.md",
+    "size": 1518,
+    "sha256": "e124dc0a0db1bb631f0eeb555cee9ccef2d8bb4c5d52f14e836aeef8107dbf0e"
+  },
+  {
+    "path": "README.md",
+    "size": 14975,
+    "sha256": "8169e29a3dcf23d7dd48b517a413a92c2d8109aa6e30db51e401cbc772648bfb"
+  },
+  {
+    "path": "REPRODUCE.md",
+    "size": 3934,
+    "sha256": "4eb94f113b30688afcc42b84c6654d07d864f3daab8b815e6d647594f031a6d6"
+  },
+  {
+    "path": "requirements.txt",
+    "size": 535,
+    "sha256": "769f98c2ced47f56ee45154c6b406f3e6a2b57cce789e32859b5784ec57cb331"
+  },
+  {
+    "path": "scheduler/scheduler_config.json",
+    "size": 485,
+    "sha256": "5895f3a167c14a967fe9ac70c64924ae5acc79799e0679fd12907e594a713cd1"
+  },
+  {
+    "path": "scripts/__init__.py",
+    "size": 49,
+    "sha256": "b79bff612ea147f36889696512ad2f99face22211fe18191c43ec70b407cdcd5"
+  },
+  {
+    "path": "scripts/benchmark.py",
+    "size": 8460,
+    "sha256": "83bdbf97778874d8d7b2a605e357940cb6a5e0f3e05bed12bd3f2d176deceb06"
+  },
+  {
+    "path": "scripts/device.py",
+    "size": 959,
+    "sha256": "aab115693753a43732b52873e11ec2704855f8631b52ddacdd7fc976f55da3cd"
+  },
+  {
+    "path": "scripts/editing_protocol.py",
+    "size": 1228,
+    "sha256": "bf144383363627fff54502538106b18c6b4be7d990a5d85d0a8f6a6182106818"
+  },
+  {
+    "path": "scripts/infer.py",
+    "size": 2842,
+    "sha256": "2c9eefe9fecea041dfa3a7dbd6d4825e6dd81a526e6d3c018bd927efea431c91"
+  },
+  {
+    "path": "scripts/integrity.py",
+    "size": 2151,
+    "sha256": "6fc5f65222e70b85d56919f712db6402c9f7c22b41694e6f8dc2b0563d9d1297"
+  },
+  {
+    "path": "scripts/model_card.py",
+    "size": 5477,
+    "sha256": "3ddd8131f9db4603184aecb87c4d36a41ba4153b2d3b9b7fdb102159cb915210"
+  },
+  {
+    "path": "scripts/provenance.py",
+    "size": 4681,
+    "sha256": "0920f7e3bd38d83713c4ebb2cb1775af97e06d760fb2b3bd520ce7d6da6aaa15"
+  },
+  {
+    "path": "scripts/quantize.py",
+    "size": 7755,
+    "sha256": "b1e47a72d59a24739a6939e12e3c9cbff0c38991fc2e52cfec1b029de878b515"
+  },
+  {
+    "path": "scripts/release.py",
+    "size": 12518,
+    "sha256": "17246624c3139e270ae5ed987fd5eb40e4bcc01d5802ab203514f93821ae5d9e"
+  },
+  {
+    "path": "scripts/report.py",
+    "size": 7953,
+    "sha256": "014757d9e7848cede3c93e9c48824df6a61cb3148bf751f6bcdf14190ae476c8"
+  },
+  {
+    "path": "scripts/runtime.py",
+    "size": 6380,
+    "sha256": "4d5e681b2d046f6993c94d5cb4f5dc39049062c5e38443f84425efbd9cec315e"
+  },
+  {
+    "path": "scripts/upload.py",
+    "size": 3403,
+    "sha256": "8b69d470fd3ba8f0ba7865eb4f533975089bef91961702240afb7c90b618e367"
+  },
+  {
+    "path": "tests/__init__.py",
+    "size": 0,
+    "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
+  },
+  {
+    "path": "tests/test_device.py",
+    "size": 1073,
+    "sha256": "8c1cc2b9800232983120bb0ca3f7d2dd4b97229ea1a8e8b9cf37a1c58c0b7f3c"
+  },
+  {
+    "path": "tests/test_integrity.py",
+    "size": 1686,
+    "sha256": "b3979412c5c675a672f3a25eb7a3c4b9178301fd2c578e9aa541393ab7263af5"
+  },
+  {
+    "path": "tests/test_sdnq_eager.py",
+    "size": 1288,
+    "sha256": "e96e42800c0a3737ce24ddef680e4a6ab97d6d89f57653519e2eaa1c3a27f114"
+  },
+  {
+    "path": "tests/test_upload_auth.py",
+    "size": 859,
+    "sha256": "47a23beb1cea4e0934425e6881f5c925607b0e57fb914f89d0d4d7e5173a4bd3"
+  },
+  {
+    "path": "text_encoder/config.json",
+    "size": 13320,
+    "sha256": "08f9c2ccf4763fa08769ebf0d5ff07e023b7849ab7fdb6be7f7f68c1c638dd15"
+  },
+  {
+    "path": "text_encoder/generation_config.json",
+    "size": 468,
+    "sha256": "013502676c64afb419718353b70249641b19f89fc8cc316177730bb957ecf9ed"
+  },
+  {
+    "path": "text_encoder/model-00001-of-00002.safetensors",
+    "size": 4986732760,
+    "sha256": "eb7c22e8c9b5597dbcee68d83f2e44f2eca29e5c942bbd22b4ff1743f293d30d"
+  },
+  {
+    "path": "text_encoder/model-00002-of-00002.safetensors",
+    "size": 2737783200,
+    "sha256": "e586e01002cc266f9794fdb7cfb1ea290a603f45a704ce08058ec15461532b4e"
+  },
+  {
+    "path": "text_encoder/model.safetensors.index.json",
+    "size": 167983,
+    "sha256": "d5c92794b905d777d3279918111a75b412805908fc1fdc773e6e681bce633cb6"
+  },
+  {
+    "path": "text_encoder-quantization.json",
+    "size": 26113,
+    "sha256": "5bffd40a1d76a7a22b384d2f765d2ed73c851d2018f47ad013aa83577cc4c3d7"
+  },
+  {
+    "path": "transformer/config.json",
+    "size": 5704,
+    "sha256": "62a3251c7145ee0d5ae271a324126b2f8548f69785fa1e3f0f4073d63704dece"
+  },
+  {
+    "path": "transformer/diffusion_pytorch_model.safetensors",
+    "size": 4365385520,
+    "sha256": "5c83319d84f2f985b289b5c69f0f98fd0a7b68a0b60b25550c36c24c33f32f5a"
+  },
+  {
+    "path": "transformer-quantization.json",
+    "size": 15583,
+    "sha256": "ad92f7d2dba5e72b47bc93c0dd06407bfa22faadc47bb75724e955df1c39c654"
+  },
+  {
+    "path": "upstream-verification.json",
+    "size": 1439,
+    "sha256": "2878036fc34cd72df293bd8432518a7a9ffbf9fe98d9ba13e2e1ed3a32f878e5"
+  },
+  {
+    "path": "vae/config.json",
+    "size": 2079,
+    "sha256": "9785d527b278cb8b210e9f48a8028d92a4190968ce7e24c8fc85d9d1b82f6ba2"
+  },
+  {
+    "path": "vae/diffusion_pytorch_model.safetensors",
+    "size": 1350989512,
+    "sha256": "a07a1b7c4ee2966a1b3bdc37de9b4f983d56937e46619f709a80b6e490675417"
+  }
+]

Notice

@@ -0,0 +1,6 @@
+Qwen is licensed under the Qwen RESEARCH LICENSE AGREEMENT, Copyright (c) 2026 Hangzhou Tongyi Laboratory Technology Co., Ltd. All Rights Reserved.
+
+Built with Qwen
+Independent derivative: Image21-INT4, by ixim / iximbox.
+Modified files: transformer and text_encoder weight shards, shard indexes and config.json files. Eligible linear weights converted to SDNQ UINT4 with SVD rank 32; floating-point exceptions are documented in component quantization reports.
+VAE, scheduler and processor retained from the pinned upstream snapshot.

PUBLISHING.md

@@ -0,0 +1,43 @@
+# 发布步骤
+
+在 `D:\Git\Qwen-Image-2.1-INT4` 打开 PowerShell。不要把 Token 贴到聊天里。
+
+打包目录：
+
+- Hugging Face：`release/huggingface`，英文模型卡，目标 `ixim/Image21-INT4`。
+- ModelScope：`release/modelscope`，中文模型卡，目标 `iximbox/Image21-INT4`。
+
+上传前：
+
+```powershell
+.venv\Scripts\python.exe -m scripts.upload huggingface --dry-run
+.venv\Scripts\python.exe -m scripts.upload modelscope --dry-run
+```
+
+## Hugging Face
+
+沿用已经登录的 ixim 缓存时，把 `HF_HOME` 指到那个目录，而不是把 Token 复制出来：
+
+```powershell
+$env:HF_HOME = "D:\Git\Qwen-Image-2.1\.cache\huggingface"
+.venv\Scripts\python.exe -m scripts.upload huggingface
+```
+
+尚未登录时，先在同一个 `HF_HOME` 下执行 `.venv\Scripts\hf.exe auth login`。
+脚本会核对账号是 `ixim`，再创建公开仓库并只上传清单里的文件。
+
+完成后查看：https://huggingface.co/ixim/Image21-INT4
+
+## ModelScope
+
+SDK 登录和网页登录是分开的。Git token 不能代替 SDK token。
+
+```powershell
+.venv\Scripts\python.exe -c "from getpass import getpass; from modelscope.hub.api import HubApi; HubApi().login(getpass('ModelScope SDK token: '))"
+.venv\Scripts\python.exe -m scripts.upload modelscope
+```
+
+脚本会核对账号是 `iximbox`。完成后查看：https://modelscope.cn/models/iximbox/Image21-INT4
+
+中断后可以再执行同一条上传命令。脚本不会删除远程文件。
+源 BF16、虚拟环境和缓存不属于发布包。

README.md

@@ -0,0 +1,302 @@
+---
+license: other
+license_name: qwen-research
+license_link: LICENSE
+base_model: Qwen/Qwen-Image-2.1
+base_model_relation: quantized
+library_name: diffusers
+pipeline_tag: text-to-image
+language:
+  - en
+  - zh
+tags:
+  - diffusers
+  - sdnq
+  - int4
+  - uint4
+  - image-generation
+  - image-editing
+  - apple-silicon
+---
+
+# Image21-INT4
+
+**Built with Qwen.** An independent 4-bit derivative of
+[Qwen-Image-2.1](https://huggingface.co/Qwen/Qwen-Image-2.1), prepared by **ixim**.
+This is a community conversion, not an official Qwen release.
+
+The stored weights are **SDNQ UINT4** (unsigned 4-bit integers) plus a rank-32 SVD
+residual. "INT4" in the repository name means this 4-bit integer checkpoint. It is
+not bitsandbytes NF4, GPTQ, or a pure integer pipeline. Quantized matmul is left
+off, so the CUDA and Apple Silicon paths dequantize with ordinary PyTorch operators.
+
+## What runs where
+
+- **CUDA, about 8GB:** group offload keeps one transformer block, or one text-encoder
+  linear layer, on the GPU. VAE encode and decode run on CPU so a 1024 activation
+  does not sit beside the denoiser cache. A 1024×1024, 40-step generation completed under a
+  7.2 GiB PyTorch allocator cap; see the measurement below. That cap was applied on
+  a 32GB RTX 5090, so it is a memory-ceiling test rather than a log from a physical
+  8GB card. The installed PyTorch build still has to support the GPU.
+- **Larger CUDA GPUs:** the loader uses model CPU offload, one pipeline component
+  at a time. The paired comparison below used this path.
+- **Apple Silicon:** the same files load on MPS. This Windows host has no Apple GPU,
+  so there is no MPS timing or memory measurement here. Unified memory has to hold
+  the packed checkpoint plus activations; an 8GB Mac is not the target. 24GB of
+  unified memory is the comfortable size for a resident load. Set
+  `IMAGE21_DTYPE=float16` if a particular MPS operator rejects bfloat16.
+
+## Quantization
+
+Base revision: `b3179ad355be050328e483a9dfdd9e60cd62adfa`.
+Eligible linear layers in the transformer and the Qwen3-VL text encoder are UINT4.
+Sensitive projections, normalization, embeddings, the vision tower, the output head,
+and the VAE stay in floating point. Exact module names are in the component
+quantization JSON files. No calibration set and no fine-tuning were used.
+
+## Installation
+
+```bash
+pip install torch torchvision
+pip install -r requirements.txt
+```
+
+On NVIDIA machines, install a CUDA wheel of PyTorch 2.10 that matches the GPU
+before `requirements.txt`. The conversion was made with torch 2.10.0+cu128.
+
+```python
+import torch
+import sys
+from huggingface_hub import snapshot_download
+
+model_dir = snapshot_download("ixim/Image21-INT4")
+sys.path.insert(0, model_dir)
+from scripts.runtime import load_pipeline
+pipe = load_pipeline(model_dir, local_files_only=True)
+image = pipe(
+    prompt='A neon sign reading "CREATE WITH LIGHT", rainy night',
+    width=1024, height=1024, num_inference_steps=40,
+    true_cfg_scale=1.0, use_kv_cache=True,
+    generator=torch.Generator("cpu").manual_seed(42),
+).images[0]
+image.save("output.png")
+```
+
+`load_pipeline` selects CUDA, MPS, or CPU. On an 8GB-class CUDA board it selects
+group offload by itself. Do not re-quantize these files at load time.
+
+For an edit, keep the edit seed different from the seed that created the source
+image. The published edit pairs use source seed 42 and edit seeds 1000042 and
+1000123. Save PNG when you need the alpha channel.
+
+## Informal evaluation
+
+### Measured comparison
+
+14 pairs, 1024×1024, 40 steps, CFG 1, KV cache on, model CPU offload, one excluded warmup. Host: NVIDIA GeForce RTX 5090.
+
+|  | BF16 | INT4 |
+|---|---|---|
+| Weight files (decimal GB) | 33.116 | 13.441 |
+| Mean call latency (s) | 26.98 | 29.40 |
+| Maximum allocated CUDA memory (GiB) | 19.08 | 9.99 |
+
+| Case | Seed | BF16 s | INT4 s | BF16 GiB | INT4 GiB | RGB MAE |
+|---|---|---|---|---|---|---|
+| chinese_text | 42 | 25.55 | 28.67 | 16.41 | 7.71 | 0.0791 |
+| chinese_text | 123 | 25.61 | 28.79 | 16.41 | 7.71 | 0.0819 |
+| composition | 42 | 25.21 | 28.72 | 16.41 | 7.71 | 0.0297 |
+| composition | 123 | 25.51 | 28.67 | 16.41 | 7.71 | 0.0594 |
+| edit | 1000042 | 38.67 | 33.81 | 19.08 | 9.99 | 0.0112 |
+| edit | 1000123 | 31.10 | 33.51 | 19.08 | 9.99 | 0.0107 |
+| english_text | 42 | 25.40 | 28.35 | 16.41 | 7.71 | 0.0839 |
+| english_text | 123 | 25.23 | 28.67 | 16.41 | 7.71 | 0.0547 |
+| portrait | 42 | 28.05 | 29.41 | 16.40 | 7.71 | 0.0528 |
+| portrait | 123 | 26.06 | 28.21 | 16.40 | 7.71 | 0.0639 |
+| rgba | 42 | 25.02 | 28.55 | 16.40 | 7.70 | 0.0444 |
+| rgba | 123 | 25.76 | 28.54 | 16.40 | 7.70 | 0.0760 |
+| texture | 42 | 25.40 | 28.98 | 16.40 | 7.70 | 0.0775 |
+| texture | 123 | 25.13 | 28.76 | 16.40 | 7.70 | 0.0489 |
+
+### Package versions
+
+| Package | Version |
+|---|---|
+| torch | 2.10.0+cu128 |
+| diffusers | 0.41.0.dev0 |
+| transformers | 5.17.0 |
+| accelerate | 1.15.0 |
+| sdnq | 0.2.7 |
+| safetensors | 0.8.0 |
+| huggingface-hub | 1.32.0 |
+| tokenizers | 0.23.2 |
+| numpy | 2.2.6 |
+| pillow | 12.2.0 |
+| psutil | 7.0.0 |
+
+### Visual inspection
+
+# Visual inspection
+
+These notes describe images I looked at. They are not a score, and they do not cover every seed. Pixel drift for all 14 pairs is in comparison.csv.
+
+portrait, seed 42: Both are waist-up photographs of an older woman in a blue wool sweater beside a window. The face, the hair, and how much of the window is in frame differ. Skin and knit texture stay photographic in both.
+
+english_text, seed 42: Both render the headline CREATE WITH LIGHT and the subtitle September 2026. The orange lamp, books, and notebook are arranged differently. I did not see an extra headline.
+
+chinese_text, seed 42: Both render 慢下来，喝杯咖啡 and 小店今日营业 above a latte. Both also add a footer of unwanted, partly illegible text, so neither follows “不添加其他文字” completely.
+
+composition, seed 42: Both show exactly three cups, red then blue then yellow, and one green apple in front of the blue cup. Cup shape and spacing differ. The INT4 cups have handles; the BF16 cups do not.
+
+texture, seed 42: Both are plausible kingfishers on a mossy branch over water. The INT4 bird faces left and the BF16 bird faces right. Feather and bark detail are visible in both.
+
+rgba, seed 42: Both are full-body green cartoon dragons with orange wings, saved as RGBA. Measured alpha runs from 0 to 255. Pixels with alpha at or below 5 are 63.5% of the BF16 image and 73.0% of the INT4 image. The BF16 dragon has a white sticker outline.
+
+edit, seed 1000042: The input is the BF16 portrait from seed 42, and the edit seed is not 42. Both change the sweater from blue to red and keep the same face, window light, and pose. This pair is not visibly oversharpened. That does not show that every edit will preserve identity.
+
+The 8GB-cap portrait (group offload, CPU VAE encode and decode) is a separate run from this paired set. On that image the woman, blue sweater, and window are all present. The same cap also completed both edit seeds, 1000042 and 1000123, at 1024 for 40 steps. Allocated peaks were 3.48 GiB and reserved peaks were 4.08 and 4.24 GiB.
+
+# 目视记录
+
+下面只记录看过的图，不是评分，也没有逐张覆盖第二个种子。14 对的像素差在 comparison.csv。
+
+portrait，种子 42：两种精度都是窗边、蓝毛衣、半身的老年女性照片。脸、头发和窗户入画的多少不同。皮肤和针织纹理都还像照片。
+
+english_text，种子 42：两种精度都写出了 CREATE WITH LIGHT 和 September 2026。橙色台灯、书和笔记本的摆放不同。没有看到多出来的主标题。
+
+chinese_text，种子 42：两种精度都写出了“慢下来，喝杯咖啡”和“小店今日营业”，中间是拿铁。两种精度的页脚都出现了多余且部分无法辨认的文字，都没有完全遵守“不添加其他文字”。
+
+composition，种子 42：两种精度都是左红、中蓝、右黄三个杯子，蓝杯前有一个青苹果。杯子形状和间距不同。INT4 的杯子有把手，BF16 没有。
+
+texture，种子 42：两种精度都是水边苔枝上的翠鸟。INT4 的鸟朝左，BF16 的鸟朝右。羽毛和树皮细节都还在。
+
+rgba，种子 42：两种精度都是全身、绿鳞、橙翼的卡通龙，文件为 RGBA。alpha 范围是 0 到 255。alpha 小于等于 5 的像素，BF16 占 63.5%，INT4 占 73.0%。BF16 的龙有一圈白色贴纸描边。
+
+edit，种子 1000042：输入是 BF16 种子 42 的肖像，编辑种子不是 42。两种精度都把蓝毛衣改成了红毛衣，脸、窗光和姿态还在。这一对没有明显过锐。这不能说明每次编辑都会保住身份。
+
+8GB 分配上限下的肖像是另一次运行（分组卸载，VAE 的编码和解码都在 CPU）。那张图里仍有人物、蓝毛衣和窗户。同一次上限设置也完成了两个编辑种子 1000042 和 1000123，1024、40 步。已分配峰值都是 3.48 GiB，保留峰值分别是 4.08 GiB 和 4.24 GiB。
+
+### 8GB CUDA allocator proof
+
+The saved INT4 checkpoint generated portrait seed 42 at 1024×1024 for 40 steps while PyTorch's caching allocator was limited to 7.2 GiB on this NVIDIA GeForce RTX 5090. Offload mode: group: one transformer block or one text-encoder leaf on GPU, VAE encode and decode on CPU. Allocated peak 1.22 GiB, reserved peak 2.37 GiB, call 84.32s after an excluded warmup. This is a ceiling test on a larger board, not a run on a physical 8GB card.
+
+![8GB cap portrait](evaluation/vram8/portrait-s42.png)
+
+
+## Paired images
+
+### chinese_text / seed 42
+
+设计一张现代咖啡馆海报，米白背景，中央一杯拉花拿铁。顶部准确写上中文大标题“慢下来，喝杯咖啡”，下方准确写上“小店今日营业”。文字清晰，留白充分，暖色摄影，不添加其他文字。
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/chinese_text-s42.png) | ![INT4](evaluation/int4/chinese_text-s42.png) |
+
+### chinese_text / seed 123
+
+设计一张现代咖啡馆海报，米白背景，中央一杯拉花拿铁。顶部准确写上中文大标题“慢下来，喝杯咖啡”，下方准确写上“小店今日营业”。文字清晰，留白充分，暖色摄影，不添加其他文字。
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/chinese_text-s123.png) | ![INT4](evaluation/int4/chinese_text-s123.png) |
+
+### composition / seed 42
+
+A studio photograph on a light gray tabletop: exactly three ceramic cups in a row, a red cup on the left, a blue cup in the middle, and a yellow cup on the right. A single green apple sits in front of the blue cup. Soft shadows, no text.
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/composition-s42.png) | ![INT4](evaluation/int4/composition-s42.png) |
+
+### composition / seed 123
+
+A studio photograph on a light gray tabletop: exactly three ceramic cups in a row, a red cup on the left, a blue cup in the middle, and a yellow cup on the right. A single green apple sits in front of the blue cup. Soft shadows, no text.
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/composition-s123.png) | ![INT4](evaluation/int4/composition-s123.png) |
+
+### edit / seed 1000042
+
+Change only the blue sweater to a red sweater. Preserve the same person, face, pose, lighting and background.
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/edit-s1000042.png) | ![INT4](evaluation/int4/edit-s1000042.png) |
+
+### edit / seed 1000123
+
+Change only the blue sweater to a red sweater. Preserve the same person, face, pose, lighting and background.
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/edit-s1000123.png) | ![INT4](evaluation/int4/edit-s1000123.png) |
+
+### english_text / seed 42
+
+A clean editorial poster with a deep blue background. Large exact headline at the top: "CREATE WITH LIGHT". Smaller exact subtitle: "September 2026". A realistic orange desk lamp occupies the lower half. Elegant balanced typography, no additional text.
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/english_text-s42.png) | ![INT4](evaluation/int4/english_text-s42.png) |
+
+### english_text / seed 123
+
+A clean editorial poster with a deep blue background. Large exact headline at the top: "CREATE WITH LIGHT". Smaller exact subtitle: "September 2026". A realistic orange desk lamp occupies the lower half. Elegant balanced typography, no additional text.
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/english_text-s123.png) | ![INT4](evaluation/int4/english_text-s123.png) |
+
+### portrait / seed 42
+
+A natural documentary portrait of an elderly woman with silver hair and freckles, wearing a blue wool sweater, standing beside a window in soft morning light. Realistic skin texture, gentle expression, waist-up composition.
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/portrait-s42.png) | ![INT4](evaluation/int4/portrait-s42.png) |
+
+### portrait / seed 123
+
+A natural documentary portrait of an elderly woman with silver hair and freckles, wearing a blue wool sweater, standing beside a window in soft morning light. Realistic skin texture, gentle expression, waist-up composition.
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/portrait-s123.png) | ![INT4](evaluation/int4/portrait-s123.png) |
+
+### rgba / seed 42
+
+This is an RGBA image with transparency. A cute cartoon dragon sticker, full body, green scales and small orange wings. The image has alpha channel and the background is transparent.
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/rgba-s42.png) | ![INT4](evaluation/int4/rgba-s42.png) |
+
+### rgba / seed 123
+
+This is an RGBA image with transparency. A cute cartoon dragon sticker, full body, green scales and small orange wings. The image has alpha channel and the background is transparent.
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/rgba-s123.png) | ![INT4](evaluation/int4/rgba-s123.png) |
+
+### texture / seed 42
+
+Macro photography of a small kingfisher perched on a mossy branch beside clear water, detailed blue feathers, droplets, natural sunlight, softly blurred forest background, realistic textures.
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/texture-s42.png) | ![INT4](evaluation/int4/texture-s42.png) |
+
+### texture / seed 123
+
+Macro photography of a small kingfisher perched on a mossy branch beside clear water, detailed blue feathers, droplets, natural sunlight, softly blurred forest background, realistic textures.
+
+| BF16 | INT4 |
+|---|---|
+| ![BF16](evaluation/bf16/texture-s123.png) | ![INT4](evaluation/int4/texture-s123.png) |
+
+
+The Qwen Research License keeps this derivative limited to non-commercial research
+and evaluation. License, Notice, and CHANGES.md are in the repository.

REPRODUCE.md

@@ -0,0 +1,89 @@
+# Image21-INT4
+
+把 Qwen-Image-2.1 做成可重新加载的 Diffusers 4-bit 衍生模型，并发布到
+Hugging Face（ixim/Image21-INT4）和 ModelScope（iximbox/Image21-INT4）。
+
+权重是 **SDNQ UINT4**，另带秩为 32 的 SVD 残差。仓库名里的 INT4 指这种 4-bit
+整数格式。量化矩阵乘法保持关闭，CUDA 和 Apple Silicon 走同一条 PyTorch eager
+反量化路径，不依赖 bitsandbytes，也不依赖 Triton。
+
+原模型使用 Qwen Research License，仅限非商业研究与评测。主名称是 **Image21-INT4**。
+
+## 运行设备
+
+加载器按机器选择路径：
+
+- CUDA 且显存不超过 10 GiB（包含 8GB 显卡）：分组卸载。Transformer 一次只把一个 block 放到 GPU，文本编码器一次只放一个线性层或嵌入层，VAE 在 CPU 上编码和解码。
+- 更大的 CUDA 显卡：一次只把一个组件放到 GPU（model CPU offload）。
+- Apple Silicon：整模加载到 MPS。统一内存要装下打包权重和激活，舒适的配置是 24GB；8GB 内存的 Mac 不是目标。这台 Windows 机器没有 Apple GPU，模型卡不会写 MPS 实测数字。
+- 没有加速器时留在 CPU。
+
+某个 MPS 算子不接受 bfloat16 时，设置 `IMAGE21_DTYPE=float16`。
+
+## 环境
+
+Windows 上复用已安装的 CUDA PyTorch：
+
+```powershell
+python -m venv .venv --system-site-packages
+.venv\Scripts\python.exe -m pip install -r requirements.txt
+```
+
+Apple Silicon 先安装 macOS 版 PyTorch，再安装 `requirements.txt`。不要安装 bitsandbytes。
+
+## 量化
+
+BF16 源权重沿用 `../Qwen-Image-2.1/models/bf16`，并按该项目已核验的 SHA256 清单检查。
+量化前不要覆盖 `models/int4`。
+
+```powershell
+.venv\Scripts\python.exe -m scripts.quantize
+.venv\Scripts\python.exe -m scripts.release annotate
+```
+
+`annotate` 写入许可证要求的修改声明。评测必须使用声明写入之后的文件。
+
+## 生成
+
+```powershell
+.venv\Scripts\python.exe -m scripts.infer --model models/int4 --prompt "一只坐在窗边的橘猫，暖色摄影" --output outputs/cat.png
+```
+
+8GB 级别的 CUDA 会自动走分组卸载。编辑示例：
+
+```powershell
+.venv\Scripts\python.exe -m scripts.infer --model models/int4 --input input.png --prompt "Change only the blue sweater to a red sweater. Preserve the same person, face, pose, lighting and background." --source-seed 42 --output outputs/edit.png
+```
+
+编辑默认种子是 1000042。它和已知的原图种子相同时会拒绝运行。
+
+## 评测
+
+先跑 BF16，再跑 INT4。编辑输入是 BF16 的 portrait-s42，编辑种子是 1000042 和 1000123，
+避免把生成原图的噪声再播一遍。两种精度使用相同的 model CPU offload。
+
+```powershell
+.venv\Scripts\python.exe -m scripts.benchmark --model ..\Qwen-Image-2.1\models\bf16 --output artifacts\eval\bf16 --offload model
+.venv\Scripts\python.exe -m scripts.benchmark --model models\int4 --output artifacts\eval\int4 --offload model
+.venv\Scripts\python.exe -m scripts.report
+```
+
+8GB 上限测试把 PyTorch 分配器限制在 7.2 GiB，并强制预算卸载。这是在更大显卡上的上限测试：
+
+```powershell
+.venv\Scripts\python.exe -m scripts.benchmark --model models\int4 --output artifacts\eval\vram8 --case portrait --seeds 42 --offload group --memory-cap-gib 7.2
+```
+
+看过图片之后把目视记录写入 `artifacts/eval/qualitative.md`，七个用例名都要出现。
+没有这份记录不能打包发布。
+
+## 发布
+
+```powershell
+.venv\Scripts\python.exe -m scripts.release stage --platform huggingface --output release\huggingface
+.venv\Scripts\python.exe -m scripts.release stage --platform modelscope --output release\modelscope
+.venv\Scripts\python.exe -m scripts.upload huggingface --dry-run
+.venv\Scripts\python.exe -m scripts.upload modelscope --dry-run
+```
+
+登录和正式上传见 [PUBLISHING.md](PUBLISHING.md)。没有远端校验记录时，不把发布说成已经完成。

benchmarks/cases.json

@@ -0,0 +1,9 @@
+[
+  {"id":"portrait","category":"portrait","prompt":"A natural documentary portrait of an elderly woman with silver hair and freckles, wearing a blue wool sweater, standing beside a window in soft morning light. Realistic skin texture, gentle expression, waist-up composition."},
+  {"id":"english_text","category":"typography_en","prompt":"A clean editorial poster with a deep blue background. Large exact headline at the top: \"CREATE WITH LIGHT\". Smaller exact subtitle: \"September 2026\". A realistic orange desk lamp occupies the lower half. Elegant balanced typography, no additional text."},
+  {"id":"chinese_text","category":"typography_zh","prompt":"设计一张现代咖啡馆海报，米白背景，中央一杯拉花拿铁。顶部准确写上中文大标题“慢下来，喝杯咖啡”，下方准确写上“小店今日营业”。文字清晰，留白充分，暖色摄影，不添加其他文字。"},
+  {"id":"composition","category":"spatial_counting","prompt":"A studio photograph on a light gray tabletop: exactly three ceramic cups in a row, a red cup on the left, a blue cup in the middle, and a yellow cup on the right. A single green apple sits in front of the blue cup. Soft shadows, no text."},
+  {"id":"texture","category":"texture","prompt":"Macro photography of a small kingfisher perched on a mossy branch beside clear water, detailed blue feathers, droplets, natural sunlight, softly blurred forest background, realistic textures."},
+  {"id":"rgba","category":"transparency","prompt":"This is an RGBA image with transparency. A cute cartoon dragon sticker, full body, green scales and small orange wings. The image has alpha channel and the background is transparent."},
+  {"id":"edit","category":"image_editing","prompt":"Change only the blue sweater to a red sweater. Preserve the same person, face, pose, lighting and background.","input":"artifacts/eval/bf16/portrait-s42.png","source_seed":42,"seeds":[1000042,1000123]}
+]

cards/huggingface.md

@@ -0,0 +1,101 @@
+---
+license: other
+license_name: qwen-research
+license_link: LICENSE
+base_model: Qwen/Qwen-Image-2.1
+base_model_relation: quantized
+library_name: diffusers
+pipeline_tag: text-to-image
+language:
+  - en
+  - zh
+tags:
+  - diffusers
+  - sdnq
+  - int4
+  - uint4
+  - image-generation
+  - image-editing
+  - apple-silicon
+---
+
+# Image21-INT4
+
+**Built with Qwen.** An independent 4-bit derivative of
+[Qwen-Image-2.1](https://huggingface.co/Qwen/Qwen-Image-2.1), prepared by **ixim**.
+This is a community conversion, not an official Qwen release.
+
+The stored weights are **SDNQ UINT4** (unsigned 4-bit integers) plus a rank-32 SVD
+residual. "INT4" in the repository name means this 4-bit integer checkpoint. It is
+not bitsandbytes NF4, GPTQ, or a pure integer pipeline. Quantized matmul is left
+off, so the CUDA and Apple Silicon paths dequantize with ordinary PyTorch operators.
+
+## What runs where
+
+- **CUDA, about 8GB:** group offload keeps one transformer block, or one text-encoder
+  linear layer, on the GPU. VAE encode and decode run on CPU so a 1024 activation
+  does not sit beside the denoiser cache. A 1024×1024, 40-step generation completed under a
+  7.2 GiB PyTorch allocator cap; see the measurement below. That cap was applied on
+  a 32GB RTX 5090, so it is a memory-ceiling test rather than a log from a physical
+  8GB card. The installed PyTorch build still has to support the GPU.
+- **Larger CUDA GPUs:** the loader uses model CPU offload, one pipeline component
+  at a time. The paired comparison below used this path.
+- **Apple Silicon:** the same files load on MPS. This Windows host has no Apple GPU,
+  so there is no MPS timing or memory measurement here. Unified memory has to hold
+  the packed checkpoint plus activations; an 8GB Mac is not the target. 24GB of
+  unified memory is the comfortable size for a resident load. Set
+  `IMAGE21_DTYPE=float16` if a particular MPS operator rejects bfloat16.
+
+## Quantization
+
+Base revision: `b3179ad355be050328e483a9dfdd9e60cd62adfa`.
+Eligible linear layers in the transformer and the Qwen3-VL text encoder are UINT4.
+Sensitive projections, normalization, embeddings, the vision tower, the output head,
+and the VAE stay in floating point. Exact module names are in the component
+quantization JSON files. No calibration set and no fine-tuning were used.
+
+## Installation
+
+```bash
+pip install torch torchvision
+pip install -r requirements.txt
+```
+
+On NVIDIA machines, install a CUDA wheel of PyTorch 2.10 that matches the GPU
+before `requirements.txt`. The conversion was made with torch 2.10.0+cu128.
+
+```python
+import torch
+import sys
+from huggingface_hub import snapshot_download
+
+model_dir = snapshot_download("ixim/Image21-INT4")
+sys.path.insert(0, model_dir)
+from scripts.runtime import load_pipeline
+pipe = load_pipeline(model_dir, local_files_only=True)
+image = pipe(
+    prompt='A neon sign reading "CREATE WITH LIGHT", rainy night',
+    width=1024, height=1024, num_inference_steps=40,
+    true_cfg_scale=1.0, use_kv_cache=True,
+    generator=torch.Generator("cpu").manual_seed(42),
+).images[0]
+image.save("output.png")
+```
+
+`load_pipeline` selects CUDA, MPS, or CPU. On an 8GB-class CUDA board it selects
+group offload by itself. Do not re-quantize these files at load time.
+
+For an edit, keep the edit seed different from the seed that created the source
+image. The published edit pairs use source seed 42 and edit seeds 1000042 and
+1000123. Save PNG when you need the alpha channel.
+
+## Informal evaluation
+
+{{DETAILS}}
+
+## Paired images
+
+{{SAMPLES}}
+
+The Qwen Research License keeps this derivative limited to non-commercial research
+and evaluation. License, Notice, and CHANGES.md are in the repository.

cards/modelscope.md

@@ -0,0 +1,92 @@
+---
+license: other
+license_name: qwen-research
+license_link: LICENSE
+base_model: Qwen/Qwen-Image-2.1
+base_model_relation: quantized
+library_name: diffusers
+pipeline_tag: text-to-image
+language:
+  - zh
+  - en
+tags:
+  - diffusers
+  - sdnq
+  - int4
+  - uint4
+  - image-generation
+  - image-editing
+  - apple-silicon
+---
+
+# Image21-INT4
+
+**Built with Qwen。** 这是 [Qwen-Image-2.1](https://www.modelscope.cn/models/Qwen/Qwen-Image-2.1)
+的独立 4-bit 衍生模型，由 **iximbox** 制作，不是 Qwen 官方发布。
+
+权重格式是 **SDNQ UINT4**（无符号 4-bit 整数）加秩为 32 的 SVD 残差。仓库名中的
+INT4 指这份 4-bit 整数权重，不是 bitsandbytes NF4，也不是 GPTQ，更不是全整数流水线。
+量化矩阵乘法保持关闭，因此 CUDA 与 Apple Silicon 都用普通 PyTorch 算子做反量化。
+
+## 能在哪里运行
+
+- **约 8GB 显存的 CUDA：** 分组卸载。Transformer 一次只把一个 block 放上 GPU，文本编码器一次只放一个线性层。VAE 的编码和解码在 CPU 上进行，避免 1024 的激活和去噪缓存叠在显存里。
+  1024×1024、40 步的一次生成是在 PyTorch 分配器被限制为 7.2 GiB 的条件下完成的，
+  数字见下方。限制加在一块 32GB 的 RTX 5090 上，所以这是显存上限测试，不是物理
+  8GB 显卡的运行日志。显卡仍须被所安装的 PyTorch 支持。
+- **更大的 CUDA 显卡：** 加载器按组件做 model CPU offload。下方的成对对比使用这条路径。
+- **Apple Silicon：** 同一套文件可加载到 MPS。制作这份模型的 Windows 主机没有 Apple GPU，
+  因此这里没有 MPS 的耗时或内存实测。统一内存需要容纳打包后的权重和激活；8GB 内存的
+  Mac 不是目标机。常驻加载比较从容的是 24GB 统一内存。若某个 MPS 算子不接受 bfloat16，
+  可设置 `IMAGE21_DTYPE=float16`。
+
+## 量化
+
+基座版本：`b3179ad355be050328e483a9dfdd9e60cd62adfa`。
+Transformer 与 Qwen3-VL 文本编码器中符合条件的线性层转为 UINT4。敏感投影、归一化、
+嵌入、视觉塔、输出头和 VAE 保持浮点。具体模块见两个组件的量化 JSON。没有校准数据集，
+也没有微调。
+
+## 安装
+
+```bash
+pip install torch torchvision
+pip install -r requirements.txt
+```
+
+NVIDIA 机器请先安装与显卡匹配的 PyTorch 2.10 CUDA 轮子，再安装 `requirements.txt`。
+本次转换使用 torch 2.10.0+cu128。
+
+```python
+import torch
+import sys
+from modelscope import snapshot_download
+
+model_dir = snapshot_download("iximbox/Image21-INT4")
+sys.path.insert(0, model_dir)
+from scripts.runtime import load_pipeline
+pipe = load_pipeline(model_dir, local_files_only=True)
+image = pipe(
+    prompt="雨夜霓虹灯，灯牌上写着 CREATE WITH LIGHT",
+    width=1024, height=1024, num_inference_steps=40,
+    true_cfg_scale=1.0, use_kv_cache=True,
+    generator=torch.Generator("cpu").manual_seed(42),
+).images[0]
+image.save("output.png")
+```
+
+`load_pipeline` 会选择 CUDA、MPS 或 CPU。8GB 级别的 CUDA 会自动使用分组卸载。
+不要在加载时再次量化这些文件。
+
+编辑时，编辑种子必须和生成原图的种子不同。已发布的编辑对使用原图种子 42，编辑种子
+1000042 和 1000123。需要透明通道时保存 PNG。
+
+## 非正式评测
+
+{{DETAILS}}
+
+## 成对图片
+
+{{SAMPLES}}
+
+Qwen Research License 将本衍生模型限制在非商业研究与评测。仓库内附有许可证、Notice 和 CHANGES.md。

conversion.json

@@ -0,0 +1,32 @@
+{
+  "base_model": "Qwen/Qwen-Image-2.1",
+  "base_revision": "b3179ad355be050328e483a9dfdd9e60cd62adfa",
+  "diffusers_commit": "80c7ed262aeffbeb43ef13ae04baeb9b84515a69",
+  "method": "sdnq",
+  "weights_dtype": "uint4",
+  "use_svd": true,
+  "svd_rank": 32,
+  "use_quantized_matmul": false,
+  "weight_files": [
+    {
+      "path": "text_encoder/model-00001-of-00002.safetensors",
+      "size": 4986732760,
+      "sha256": "eb7c22e8c9b5597dbcee68d83f2e44f2eca29e5c942bbd22b4ff1743f293d30d"
+    },
+    {
+      "path": "text_encoder/model-00002-of-00002.safetensors",
+      "size": 2737783200,
+      "sha256": "e586e01002cc266f9794fdb7cfb1ea290a603f45a704ce08058ec15461532b4e"
+    },
+    {
+      "path": "transformer/diffusion_pytorch_model.safetensors",
+      "size": 4365385520,
+      "sha256": "5c83319d84f2f985b289b5c69f0f98fd0a7b68a0b60b25550c36c24c33f32f5a"
+    },
+    {
+      "path": "vae/diffusion_pytorch_model.safetensors",
+      "size": 1350989512,
+      "sha256": "a07a1b7c4ee2966a1b3bdc37de9b4f983d56937e46619f709a80b6e490675417"
+    }
+  ]
+}

evaluation/bf16/environment.json

@@ -0,0 +1,181 @@
+{
+  "python": "3.13.5",
+  "platform": "Windows-11-10.0.26200-SP0",
+  "gpu": "NVIDIA GeForce RTX 5090",
+  "cuda": "12.8",
+  "mps": false,
+  "system_ram_bytes": 274298757120,
+  "packages": {
+    "torch": "2.10.0+cu128",
+    "diffusers": "0.41.0.dev0",
+    "transformers": "5.17.0",
+    "accelerate": "1.15.0",
+    "sdnq": "0.2.7",
+    "safetensors": "0.8.0",
+    "huggingface-hub": "1.32.0",
+    "tokenizers": "0.23.2",
+    "numpy": "2.2.6",
+    "pillow": "12.2.0",
+    "psutil": "7.0.0"
+  },
+  "pid": 30256,
+  "before_load_memory": {
+    "allocated_bytes": 0,
+    "reserved_bytes": 0,
+    "cuda_free_bytes": 32496418816,
+    "cuda_total_bytes": 34162016256
+  },
+  "memory_cap": null,
+  "benchmark_sha256": "83bdbf97778874d8d7b2a605e357940cb6a5e0f3e05bed12bd3f2d176deceb06",
+  "runtime_helper_sha256": "81ea21948d27f61ba210b8fc44487b057200b741df0ab8c587f60d9cec02f6f9",
+  "device_helper_sha256": "aab115693753a43732b52873e11ec2704855f8631b52ddacdd7fc976f55da3cd",
+  "gpu_before": "name, driver_version, memory.used [MiB], memory.free [MiB]\nNVIDIA GeForce RTX 5090, 616.92, 3030 MiB, 29158 MiB\n",
+  "gpu_processes_before": "pid, process_name, used_gpu_memory [MiB]\n14108, C:\\Windows\\SystemApps\\MicrosoftWindows.Client.CBS_cw5n1h2txyewy\\CrossDeviceResume.exe, [N/A]\n13756, C:\\Windows\\explorer.exe, [N/A]\n16592, C:\\Windows\\SystemApps\\Microsoft.Windows.StartMenuExperienceHost_cw5n1h2txyewy\\StartMenuExperienceHost.exe, [N/A]\n16496, C:\\Windows\\SystemApps\\MicrosoftWindows.Client.CBS_cw5n1h2txyewy\\SearchHost.exe, [N/A]\n20388, C:\\Windows\\SystemApps\\Microsoft.LockApp_cw5n1h2txyewy\\LockApp.exe, [N/A]\n19856, C:\\Program Files (x86)\\Microsoft\\EdgeWebView\\Application\\153.0.4234.48\\msedgewebview2.exe, [N/A]\n21864, C:\\Program Files\\NVIDIA Corporation\\NVIDIA App\\CEF\\NVIDIA Overlay.exe, [N/A]\n15368, C:\\Program Files\\NVIDIA Corporation\\NVIDIA App\\CEF\\NVIDIA Overlay.exe, [N/A]\n23204, C:\\Windows\\SystemApps\\MicrosoftWindows.Client.CBS_cw5n1h2txyewy\\TextInputHost.exe, [N/A]\n20596, C:\\Program Files\\LogiOptionsPlus\\logioptionsplus_agent.exe, [N/A]\n16776, C:\\Program Files\\WindowsApps\\Microsoft.YourPhone_1.26072.255.0_x64__8wekyb3d8bbwe\\PhoneExperienceHost.exe, [N/A]\n29352, C:\\Program Files\\WindowsApps\\Microsoft.6365217CE6EB4_102.2504.16004.0_x64__8wekyb3d8bbwe\\MicrosoftSecurityApp\\MicrosoftSecurityApp.exe, [N/A]\n5036, C:\\Program Files\\Intel\\Intel Graphics Software\\IntelGraphicsSoftware.exe, [N/A]\n12064, C:\\Program Files (x86)\\Microsoft\\EdgeWebView\\Application\\153.0.4234.48\\msedgewebview2.exe, [N/A]\n28020, C:\\Windows\\System32\\ApplicationFrameHost.exe, [N/A]\n30732, C:\\Windows\\ImmersiveControlPanel\\SystemSettings.exe, [N/A]\n30700, C:\\Users\\Admin\\AppData\\Local\\Programs\\MiniMark\\Runtime\\WebView2\\152.0.4191.53\\msedgewebview2.exe, [N/A]\n34240, C:\\Users\\Admin\\AppData\\Local\\Programs\\MiniMark\\MiniMark.Windows.exe, [N/A]\n19236, C:\\Windows\\SystemApps\\ShellExperienceHost_cw5n1h2txyewy\\ShellExperienceHost.exe, [N/A]\n3388, C:\\Program Files (x86)\\Microsoft\\Edge\\Application\\msedge.exe, [N/A]\n23864, C:\\Users\\Admin\\AppData\\Local\\Programs\\WorkBuddy\\WorkBuddy.exe, [N/A]\n30016, C:\\Program Files (x86)\\Microsoft\\Edge\\Application\\msedge.exe, [N/A]\n15804, C:\\Program Files\\WindowsApps\\Microsoft.WindowsTerminal_1.24.11911.0_x64__8wekyb3d8bbwe\\WindowsTerminal.exe, [N/A]\n28420, C:\\Windows\\System32\\ShellHost.exe, [N/A]\n30256, C:\\Python313\\python.exe, [N/A]\n",
+  "model_identity": {
+    "kind": "bf16",
+    "base_revision": "b3179ad355be050328e483a9dfdd9e60cd62adfa",
+    "files": [
+      {
+        "path": "model_index.json",
+        "size": 447,
+        "sha256": "cf1ecd104ea090855d60d8cec0895c1e9d6ee41b2f87e231b678beecd1cf7809"
+      },
+      {
+        "path": "processor/added_tokens.json",
+        "size": 707,
+        "sha256": "c0284b582e14987fbd3d5a2cb2bd139084371ed9acbae488829a1c900833c680"
+      },
+      {
+        "path": "processor/chat_template.jinja",
+        "size": 5292,
+        "sha256": "3636d0f0bd6bef02654cdffdc447b79cb2cef8ab02cc75267345946291a489e4"
+      },
+      {
+        "path": "processor/merges.txt",
+        "size": 1671853,
+        "sha256": "8831e4f1a044471340f7c0a83d7bd71306a5b867e95fd870f74d0c5308a904d5"
+      },
+      {
+        "path": "processor/preprocessor_config.json",
+        "size": 782,
+        "sha256": "93585062a80db5e8ca038efc7726a3e6411d9db948472d81d63c6303993be8c5"
+      },
+      {
+        "path": "processor/special_tokens_map.json",
+        "size": 613,
+        "sha256": "76862e765266b85aa9459767e33cbaf13970f327a0e88d1c65846c2ddd3a1ecd"
+      },
+      {
+        "path": "processor/tokenizer.json",
+        "size": 11422654,
+        "sha256": "aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4"
+      },
+      {
+        "path": "processor/tokenizer_config.json",
+        "size": 5445,
+        "sha256": "81ec7bb9530159b326c0bef1d0b6c33d392090524014ea3f0123a3c1eb9c2af5"
+      },
+      {
+        "path": "processor/video_preprocessor_config.json",
+        "size": 817,
+        "sha256": "59c5c9eb52182eb14c06ffb10ca9effd29adce5f238a95de23ca14a38dbd2cb1"
+      },
+      {
+        "path": "processor/vocab.json",
+        "size": 2776833,
+        "sha256": "ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910"
+      },
+      {
+        "path": "scheduler/scheduler_config.json",
+        "size": 485,
+        "sha256": "5895f3a167c14a967fe9ac70c64924ae5acc79799e0679fd12907e594a713cd1"
+      },
+      {
+        "path": "text_encoder/config.json",
+        "size": 1517,
+        "sha256": "6001331949f7f86ab6d791a80b12246d4f8da5159fa08c3e62368a39f36d63a9"
+      },
+      {
+        "path": "text_encoder/generation_config.json",
+        "size": 213,
+        "sha256": "4d9818c3d27895c0058828a5f68bc7a4de80c3ae1bcdac90936ce24178063f59"
+      },
+      {
+        "path": "text_encoder/model-00001-of-00004.safetensors",
+        "size": 4998056552,
+        "sha256": "dde00291b5f7fb92013895310a3da0ddba78674df9f10d505d375243dc01fc6f"
+      },
+      {
+        "path": "text_encoder/model-00002-of-00004.safetensors",
+        "size": 4915962464,
+        "sha256": "9047faccc0a6d98496a52d55f27be1c94a9c259d1e283fbea0128d054a948d42"
+      },
+      {
+        "path": "text_encoder/model-00003-of-00004.safetensors",
+        "size": 4915962496,
+        "sha256": "8c54187654c0176b73ae73785bf791dc9a14c9df7fb4310083a09d42048cb57e"
+      },
+      {
+        "path": "text_encoder/model-00004-of-00004.safetensors",
+        "size": 2704357976,
+        "sha256": "5311532aaaeae3259eb6a7b2c600636be1159adf7ded35f53579f7d0e7d43cdd"
+      },
+      {
+        "path": "text_encoder/model.safetensors.index.json",
+        "size": 67795,
+        "sha256": "859146bff8db1d617fca8b457a5223edff963bb70076b3645f2189fe0540a147"
+      },
+      {
+        "path": "transformer/config.json",
+        "size": 370,
+        "sha256": "56ae3281c4e6c2d1aa3658252d187488071815fd79bef15808bb0205fc1a2241"
+      },
+      {
+        "path": "transformer/diffusion_pytorch_model-00001-of-00002.safetensors",
+        "size": 9968332504,
+        "sha256": "9e6bc2d641e67bf277895ea8777141044a38f3edb7101bc469b2961dd7c36b4b"
+      },
+      {
+        "path": "transformer/diffusion_pytorch_model-00002-of-00002.safetensors",
+        "size": 4261951904,
+        "sha256": "3aaf234dcbe128530479735854a346b5e3e66283b7c11db56f836bbd1c13ebaa"
+      },
+      {
+        "path": "transformer/diffusion_pytorch_model.safetensors.index.json",
+        "size": 30283,
+        "sha256": "17987f6623b1c814d0ef55a137d99142b7b3b040eb1bf241b5575dd35af803a2"
+      },
+      {
+        "path": "vae/config.json",
+        "size": 2079,
+        "sha256": "9785d527b278cb8b210e9f48a8028d92a4190968ce7e24c8fc85d9d1b82f6ba2"
+      },
+      {
+        "path": "vae/diffusion_pytorch_model.safetensors",
+        "size": 1350989512,
+        "sha256": "a07a1b7c4ee2966a1b3bdc37de9b4f983d56937e46619f709a80b6e490675417"
+      }
+    ],
+    "fingerprint": "74c96a68092fdb1f3eed2063bcfba8fd99380ada6234e405e646681f35fb6984"
+  },
+  "model": "D:\\Git\\Qwen-Image-2.1\\models\\bf16",
+  "load_seconds": 17.745139499995275,
+  "load_peak_allocated_bytes": 8704,
+  "offload": "model",
+  "requested_offload": "model",
+  "runtime_policy": {
+    "offload": "model",
+    "dtype": "bfloat16",
+    "use_stream": false,
+    "device": "cuda"
+  },
+  "warmup": true,
+  "after_load_memory": {
+    "allocated_bytes": 0,
+    "reserved_bytes": 0,
+    "cuda_free_bytes": 32494321664,
+    "cuda_total_bytes": 34162016256
+  },
+  "generator_device": "cpu",
+  "cases_sha256": "c0bc77b0aa28c9f11442c30cafd07d31bca72a179c7b759fb2a5ed9e328896c8"
+}

evaluation/bf16/records.jsonl

@@ -0,0 +1,14 @@
+{"case_id": "portrait", "category": "portrait", "prompt": "A natural documentary portrait of an elderly woman with silver hair and freckles, wearing a blue wool sweater, standing beside a window in soft morning light. Realistic skin texture, gentle expression, waist-up composition.", "seed": 42, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 23068672, "cuda_free_bytes": 32322355200, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 54525952, "cuda_free_bytes": 32290897920, "cuda_total_bytes": 34162016256}, "seconds": 28.04724630000419, "peak_allocated_bytes": 17611056128, "peak_reserved_bytes": 17800626176, "process_rss_after_bytes": 42548174848, "image": "portrait-s42.png", "image_sha256": "01e37e178fb0d35d5c706b7edd76bc1a6c5b50ec3536901a467fd45295d097aa", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "portrait", "category": "portrait", "prompt": "A natural documentary portrait of an elderly woman with silver hair and freckles, wearing a blue wool sweater, standing beside a window in soft morning light. Realistic skin texture, gentle expression, waist-up composition.", "seed": 123, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 23068672, "cuda_free_bytes": 32322355200, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 54525952, "cuda_free_bytes": 32290897920, "cuda_total_bytes": 34162016256}, "seconds": 26.058282700003474, "peak_allocated_bytes": 17611056128, "peak_reserved_bytes": 17800626176, "process_rss_after_bytes": 45176520704, "image": "portrait-s123.png", "image_sha256": "cff376e50c41c4aee2b9e4aabc502f4791774f6ad285d8a7673a61ebb1fa650d", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "english_text", "category": "typography_en", "prompt": "A clean editorial poster with a deep blue background. Large exact headline at the top: \"CREATE WITH LIGHT\". Smaller exact subtitle: \"September 2026\". A realistic orange desk lamp occupies the lower half. Elegant balanced typography, no additional text.", "seed": 42, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 23068672, "cuda_free_bytes": 32322355200, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 62914560, "cuda_free_bytes": 32282509312, "cuda_total_bytes": 34162016256}, "seconds": 25.40485580000677, "peak_allocated_bytes": 17619124224, "peak_reserved_bytes": 17806917632, "process_rss_after_bytes": 46491959296, "image": "english_text-s42.png", "image_sha256": "ba31b0c8a312e96408096566013c667b3a32bd7720749ca47120960b17b41ca6", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "english_text", "category": "typography_en", "prompt": "A clean editorial poster with a deep blue background. Large exact headline at the top: \"CREATE WITH LIGHT\". Smaller exact subtitle: \"September 2026\". A realistic orange desk lamp occupies the lower half. Elegant balanced typography, no additional text.", "seed": 123, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 23068672, "cuda_free_bytes": 32322355200, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 62914560, "cuda_free_bytes": 32282509312, "cuda_total_bytes": 34162016256}, "seconds": 25.2340533999959, "peak_allocated_bytes": 17619124224, "peak_reserved_bytes": 17806917632, "process_rss_after_bytes": 47513837568, "image": "english_text-s123.png", "image_sha256": "cf5e76dee9f0e403ebdf24a64e1fcd604ee2b6c8e645a032fba2807acc34adf2", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "chinese_text", "category": "typography_zh", "prompt": "设计一张现代咖啡馆海报，米白背景，中央一杯拉花拿铁。顶部准确写上中文大标题“慢下来，喝杯咖啡”，下方准确写上“小店今日营业”。文字清晰，留白充分，暖色摄影，不添加其他文字。", "seed": 42, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 23068672, "cuda_free_bytes": 32322355200, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 73400320, "cuda_free_bytes": 32272023552, "cuda_total_bytes": 34162016256}, "seconds": 25.548122699998203, "peak_allocated_bytes": 17623597568, "peak_reserved_bytes": 17813209088, "process_rss_after_bytes": 48606068736, "image": "chinese_text-s42.png", "image_sha256": "7082cb71d99705bc42e27e2c206b13e3beb9e62cfd6304cfabe203b82641692c", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "chinese_text", "category": "typography_zh", "prompt": "设计一张现代咖啡馆海报，米白背景，中央一杯拉花拿铁。顶部准确写上中文大标题“慢下来，喝杯咖啡”，下方准确写上“小店今日营业”。文字清晰，留白充分，暖色摄影，不添加其他文字。", "seed": 123, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 23068672, "cuda_free_bytes": 32322355200, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 73400320, "cuda_free_bytes": 32272023552, "cuda_total_bytes": 34162016256}, "seconds": 25.60944980000204, "peak_allocated_bytes": 17623597568, "peak_reserved_bytes": 17813209088, "process_rss_after_bytes": 49343434752, "image": "chinese_text-s123.png", "image_sha256": "eae74256d7ed1b3519d39c7d58961cf7a937224d19a3c3239ca7eba4178edf49", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "composition", "category": "spatial_counting", "prompt": "A studio photograph on a light gray tabletop: exactly three ceramic cups in a row, a red cup on the left, a blue cup in the middle, and a yellow cup on the right. A single green apple sits in front of the blue cup. Soft shadows, no text.", "seed": 42, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 23068672, "cuda_free_bytes": 32322355200, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 73400320, "cuda_free_bytes": 32272023552, "cuda_total_bytes": 34162016256}, "seconds": 25.20887319999747, "peak_allocated_bytes": 17621962752, "peak_reserved_bytes": 17813209088, "process_rss_after_bytes": 49948581888, "image": "composition-s42.png", "image_sha256": "91e967685fbbd79ba13daece22a0f72c863cb518db88d5d2a6f9aecfc5f292b1", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "composition", "category": "spatial_counting", "prompt": "A studio photograph on a light gray tabletop: exactly three ceramic cups in a row, a red cup on the left, a blue cup in the middle, and a yellow cup on the right. A single green apple sits in front of the blue cup. Soft shadows, no text.", "seed": 123, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 23068672, "cuda_free_bytes": 32322355200, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 73400320, "cuda_free_bytes": 32272023552, "cuda_total_bytes": 34162016256}, "seconds": 25.510518499999307, "peak_allocated_bytes": 17621962752, "peak_reserved_bytes": 17813209088, "process_rss_after_bytes": 50194882560, "image": "composition-s123.png", "image_sha256": "bc11612f260acd8af0d834f3b28c752cefd0437d24c9a01dae0d46647bdf29ae", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "texture", "category": "texture", "prompt": "Macro photography of a small kingfisher perched on a mossy branch beside clear water, detailed blue feathers, droplets, natural sunlight, softly blurred forest background, realistic textures.", "seed": 42, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 23068672, "cuda_free_bytes": 32322355200, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 54525952, "cuda_free_bytes": 32290897920, "cuda_total_bytes": 34162016256}, "seconds": 25.40272179999738, "peak_allocated_bytes": 17607326208, "peak_reserved_bytes": 17798529024, "process_rss_after_bytes": 50262564864, "image": "texture-s42.png", "image_sha256": "40a0f73bb407a394fa9e8a3f6922455df716b9725ad07437fa10c1daca2c2095", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "texture", "category": "texture", "prompt": "Macro photography of a small kingfisher perched on a mossy branch beside clear water, detailed blue feathers, droplets, natural sunlight, softly blurred forest background, realistic textures.", "seed": 123, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 25165824, "cuda_free_bytes": 32320258048, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 54525952, "cuda_free_bytes": 32290897920, "cuda_total_bytes": 34162016256}, "seconds": 25.126798500001314, "peak_allocated_bytes": 17607326208, "peak_reserved_bytes": 17798529024, "process_rss_after_bytes": 50264162304, "image": "texture-s123.png", "image_sha256": "18c4403f388d0e03447473c03497141169f333fddeaeb40227163dc3b9d68d3e", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "rgba", "category": "transparency", "prompt": "This is an RGBA image with transparency. A cute cartoon dragon sticker, full body, green scales and small orange wings. The image has alpha channel and the background is transparent.", "seed": 42, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 25165824, "cuda_free_bytes": 32320258048, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 54525952, "cuda_free_bytes": 32290897920, "cuda_total_bytes": 34162016256}, "seconds": 25.024158100000932, "peak_allocated_bytes": 17604822528, "peak_reserved_bytes": 17798529024, "process_rss_after_bytes": 50352259072, "image": "rgba-s42.png", "image_sha256": "6f5dfea6ba02e41cbf20617b6e863e1adfa139d79fad2ca5c64dfc815f591d82", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "rgba", "category": "transparency", "prompt": "This is an RGBA image with transparency. A cute cartoon dragon sticker, full body, green scales and small orange wings. The image has alpha channel and the background is transparent.", "seed": 123, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 25165824, "cuda_free_bytes": 32320258048, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 54525952, "cuda_free_bytes": 32290897920, "cuda_total_bytes": 34162016256}, "seconds": 25.75669769999513, "peak_allocated_bytes": 17604822528, "peak_reserved_bytes": 17798529024, "process_rss_after_bytes": 50135724032, "image": "rgba-s123.png", "image_sha256": "6f1abbb1a64ac553fc576a0250d370ab5bb7e6d5e6372f42f67045a5bb0951a6", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "edit", "category": "image_editing", "prompt": "Change only the blue sweater to a red sweater. Preserve the same person, face, pose, lighting and background.", "seed": 1000042, "source_seed": 42, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": "01e37e178fb0d35d5c706b7edd76bc1a6c5b50ec3536901a467fd45295d097aa", "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 25165824, "cuda_free_bytes": 32320258048, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 14288384, "reserved_bytes": 2701131776, "cuda_free_bytes": 29644292096, "cuda_total_bytes": 34162016256}, "seconds": 38.66664409999794, "peak_allocated_bytes": 20488328192, "peak_reserved_bytes": 21506293760, "process_rss_after_bytes": 50869055488, "image": "edit-s1000042.png", "image_sha256": "a846f593fd08c74231fa6154654dc089668c2e4f2359b6e65f5d136cbe8fc493", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "edit", "category": "image_editing", "prompt": "Change only the blue sweater to a red sweater. Preserve the same person, face, pose, lighting and background.", "seed": 1000123, "source_seed": 42, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": "01e37e178fb0d35d5c706b7edd76bc1a6c5b50ec3536901a467fd45295d097aa", "before_memory": {"allocated_bytes": 14288384, "reserved_bytes": 27262976, "cuda_free_bytes": 32318160896, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 14288384, "reserved_bytes": 2701131776, "cuda_free_bytes": 29644292096, "cuda_total_bytes": 34162016256}, "seconds": 31.102813400008017, "peak_allocated_bytes": 20488328192, "peak_reserved_bytes": 21506293760, "process_rss_after_bytes": 50333995008, "image": "edit-s1000123.png", "image_sha256": "e46d6f1316e09816be7d980a65277e53fbb5ba25578e2f6dfcd7e890d1b14b32", "image_mode": "RGBA", "actual_size": [1024, 1024]}

evaluation/comparison.csv

@@ -0,0 +1,15 @@
+case_id,seed,bf16_seconds,int4_seconds,bf16_peak_allocated_gib,int4_peak_allocated_gib,rgb_mae_white,rgb_psnr_white_db,alpha_mae
+portrait,42,28.04724630000419,29.40627289999975,16.40157413482666,7.7064409255981445,0.05282028019428253,22.529744751933244,9.150783444056287e-05
+portrait,123,26.058282700003474,28.207035100000212,16.40157413482666,7.7064409255981445,0.06386926025152206,21.04555004835983,0.0003564759681466967
+english_text,42,25.40485580000677,28.350070699991193,16.409088134765625,7.710947036743164,0.08393288403749466,13.985232485933237,1.007154605758842e-05
+english_text,123,25.2340533999959,28.674008000001777,16.409088134765625,7.710947036743164,0.05471445992588997,15.336678147708,1.027350481308531e-05
+chinese_text,42,25.548122699998203,28.668083800002933,16.413254261016846,7.713199615478516,0.07913437485694885,14.545009708680794,0.00011114605877082795
+chinese_text,123,25.60944980000204,28.79211700000451,16.413254261016846,7.713199615478516,0.08193721622228622,14.625667975050582,0.00010120537626789883
+composition,42,25.20887319999747,28.723114799999166,16.411731719970703,7.712749481201172,0.029696526005864143,21.84438550008862,3.587311221053824e-05
+composition,123,25.510518499999307,28.66992489999393,16.411731719970703,7.712749481201172,0.05939028039574623,18.007705024410267,8.58269413583912e-05
+texture,42,25.40272179999738,28.9771555000043,16.39810037612915,7.7037353515625,0.07745420187711716,17.187610768167794,0.00017304887296631932
+texture,123,25.126798500001314,28.759817099999054,16.39810037612915,7.7037353515625,0.04893387854099274,22.306011733308672,8.507524034939706e-05
+rgba,42,25.024158100000932,28.55306270001165,16.395768642425537,7.702834129333496,0.04440908506512642,16.41620600119147,0.0966198667883873
+rgba,123,25.75669769999513,28.541914399989764,16.395768642425537,7.702834129333496,0.07601191848516464,14.108690165177793,0.06474930047988892
+edit,1000042,38.66664409999794,33.80973500000255,19.081242561340332,9.985671043395996,0.011213180609047413,36.34769504163552,3.535326322889887e-05
+edit,1000123,31.102813400008017,33.51434729999164,19.081242561340332,9.985671043395996,0.010669182986021042,36.50090754042482,3.2294025004375726e-05

evaluation/comparison.html

@@ -0,0 +1,17 @@
+<!doctype html><meta charset="utf-8"><title>BF16 / INT4</title>
+<style>body{font:16px system-ui;margin:32px;max-width:1500px}.pair{display:grid;grid-template-columns:1fr 1fr;gap:20px}img{width:100%;background:repeating-conic-gradient(#ddd 0 25%,#fff 0 50%) 0/24px 24px}pre{white-space:pre-wrap}</style>
+<h1>BF16 / INT4</h1><p>Left: BF16. Right: saved INT4. Same settings and seeds.</p>
+<section><h2>portrait seed 42</h2><pre>A natural documentary portrait of an elderly woman with silver hair and freckles, wearing a blue wool sweater, standing beside a window in soft morning light. Realistic skin texture, gentle expression, waist-up composition.</pre><div class="pair"><a href="bf16/portrait-s42.png"><img src="bf16/portrait-s42.png" alt="BF16"></a><a href="int4/portrait-s42.png"><img src="int4/portrait-s42.png" alt="INT4"></a></div></section>
+<section><h2>portrait seed 123</h2><pre>A natural documentary portrait of an elderly woman with silver hair and freckles, wearing a blue wool sweater, standing beside a window in soft morning light. Realistic skin texture, gentle expression, waist-up composition.</pre><div class="pair"><a href="bf16/portrait-s123.png"><img src="bf16/portrait-s123.png" alt="BF16"></a><a href="int4/portrait-s123.png"><img src="int4/portrait-s123.png" alt="INT4"></a></div></section>
+<section><h2>english_text seed 42</h2><pre>A clean editorial poster with a deep blue background. Large exact headline at the top: &quot;CREATE WITH LIGHT&quot;. Smaller exact subtitle: &quot;September 2026&quot;. A realistic orange desk lamp occupies the lower half. Elegant balanced typography, no additional text.</pre><div class="pair"><a href="bf16/english_text-s42.png"><img src="bf16/english_text-s42.png" alt="BF16"></a><a href="int4/english_text-s42.png"><img src="int4/english_text-s42.png" alt="INT4"></a></div></section>
+<section><h2>english_text seed 123</h2><pre>A clean editorial poster with a deep blue background. Large exact headline at the top: &quot;CREATE WITH LIGHT&quot;. Smaller exact subtitle: &quot;September 2026&quot;. A realistic orange desk lamp occupies the lower half. Elegant balanced typography, no additional text.</pre><div class="pair"><a href="bf16/english_text-s123.png"><img src="bf16/english_text-s123.png" alt="BF16"></a><a href="int4/english_text-s123.png"><img src="int4/english_text-s123.png" alt="INT4"></a></div></section>
+<section><h2>chinese_text seed 42</h2><pre>设计一张现代咖啡馆海报，米白背景，中央一杯拉花拿铁。顶部准确写上中文大标题“慢下来，喝杯咖啡”，下方准确写上“小店今日营业”。文字清晰，留白充分，暖色摄影，不添加其他文字。</pre><div class="pair"><a href="bf16/chinese_text-s42.png"><img src="bf16/chinese_text-s42.png" alt="BF16"></a><a href="int4/chinese_text-s42.png"><img src="int4/chinese_text-s42.png" alt="INT4"></a></div></section>
+<section><h2>chinese_text seed 123</h2><pre>设计一张现代咖啡馆海报，米白背景，中央一杯拉花拿铁。顶部准确写上中文大标题“慢下来，喝杯咖啡”，下方准确写上“小店今日营业”。文字清晰，留白充分，暖色摄影，不添加其他文字。</pre><div class="pair"><a href="bf16/chinese_text-s123.png"><img src="bf16/chinese_text-s123.png" alt="BF16"></a><a href="int4/chinese_text-s123.png"><img src="int4/chinese_text-s123.png" alt="INT4"></a></div></section>
+<section><h2>composition seed 42</h2><pre>A studio photograph on a light gray tabletop: exactly three ceramic cups in a row, a red cup on the left, a blue cup in the middle, and a yellow cup on the right. A single green apple sits in front of the blue cup. Soft shadows, no text.</pre><div class="pair"><a href="bf16/composition-s42.png"><img src="bf16/composition-s42.png" alt="BF16"></a><a href="int4/composition-s42.png"><img src="int4/composition-s42.png" alt="INT4"></a></div></section>
+<section><h2>composition seed 123</h2><pre>A studio photograph on a light gray tabletop: exactly three ceramic cups in a row, a red cup on the left, a blue cup in the middle, and a yellow cup on the right. A single green apple sits in front of the blue cup. Soft shadows, no text.</pre><div class="pair"><a href="bf16/composition-s123.png"><img src="bf16/composition-s123.png" alt="BF16"></a><a href="int4/composition-s123.png"><img src="int4/composition-s123.png" alt="INT4"></a></div></section>
+<section><h2>texture seed 42</h2><pre>Macro photography of a small kingfisher perched on a mossy branch beside clear water, detailed blue feathers, droplets, natural sunlight, softly blurred forest background, realistic textures.</pre><div class="pair"><a href="bf16/texture-s42.png"><img src="bf16/texture-s42.png" alt="BF16"></a><a href="int4/texture-s42.png"><img src="int4/texture-s42.png" alt="INT4"></a></div></section>
+<section><h2>texture seed 123</h2><pre>Macro photography of a small kingfisher perched on a mossy branch beside clear water, detailed blue feathers, droplets, natural sunlight, softly blurred forest background, realistic textures.</pre><div class="pair"><a href="bf16/texture-s123.png"><img src="bf16/texture-s123.png" alt="BF16"></a><a href="int4/texture-s123.png"><img src="int4/texture-s123.png" alt="INT4"></a></div></section>
+<section><h2>rgba seed 42</h2><pre>This is an RGBA image with transparency. A cute cartoon dragon sticker, full body, green scales and small orange wings. The image has alpha channel and the background is transparent.</pre><div class="pair"><a href="bf16/rgba-s42.png"><img src="bf16/rgba-s42.png" alt="BF16"></a><a href="int4/rgba-s42.png"><img src="int4/rgba-s42.png" alt="INT4"></a></div></section>
+<section><h2>rgba seed 123</h2><pre>This is an RGBA image with transparency. A cute cartoon dragon sticker, full body, green scales and small orange wings. The image has alpha channel and the background is transparent.</pre><div class="pair"><a href="bf16/rgba-s123.png"><img src="bf16/rgba-s123.png" alt="BF16"></a><a href="int4/rgba-s123.png"><img src="int4/rgba-s123.png" alt="INT4"></a></div></section>
+<section><h2>edit seed 1000042</h2><pre>Change only the blue sweater to a red sweater. Preserve the same person, face, pose, lighting and background.</pre><div class="pair"><a href="bf16/edit-s1000042.png"><img src="bf16/edit-s1000042.png" alt="BF16"></a><a href="int4/edit-s1000042.png"><img src="int4/edit-s1000042.png" alt="INT4"></a></div></section>
+<section><h2>edit seed 1000123</h2><pre>Change only the blue sweater to a red sweater. Preserve the same person, face, pose, lighting and background.</pre><div class="pair"><a href="bf16/edit-s1000123.png"><img src="bf16/edit-s1000123.png" alt="BF16"></a><a href="int4/edit-s1000123.png"><img src="int4/edit-s1000123.png" alt="INT4"></a></div></section>

evaluation/int4/environment.json

@@ -0,0 +1,161 @@
+{
+  "python": "3.13.5",
+  "platform": "Windows-11-10.0.26200-SP0",
+  "gpu": "NVIDIA GeForce RTX 5090",
+  "cuda": "12.8",
+  "mps": false,
+  "system_ram_bytes": 274298757120,
+  "packages": {
+    "torch": "2.10.0+cu128",
+    "diffusers": "0.41.0.dev0",
+    "transformers": "5.17.0",
+    "accelerate": "1.15.0",
+    "sdnq": "0.2.7",
+    "safetensors": "0.8.0",
+    "huggingface-hub": "1.32.0",
+    "tokenizers": "0.23.2",
+    "numpy": "2.2.6",
+    "pillow": "12.2.0",
+    "psutil": "7.0.0"
+  },
+  "pid": 4580,
+  "before_load_memory": {
+    "allocated_bytes": 0,
+    "reserved_bytes": 0,
+    "cuda_free_bytes": 32496418816,
+    "cuda_total_bytes": 34162016256
+  },
+  "memory_cap": null,
+  "benchmark_sha256": "83bdbf97778874d8d7b2a605e357940cb6a5e0f3e05bed12bd3f2d176deceb06",
+  "runtime_helper_sha256": "81ea21948d27f61ba210b8fc44487b057200b741df0ab8c587f60d9cec02f6f9",
+  "device_helper_sha256": "aab115693753a43732b52873e11ec2704855f8631b52ddacdd7fc976f55da3cd",
+  "gpu_before": "name, driver_version, memory.used [MiB], memory.free [MiB]\nNVIDIA GeForce RTX 5090, 616.92, 3001 MiB, 29187 MiB\n",
+  "gpu_processes_before": "pid, process_name, used_gpu_memory [MiB]\n14108, C:\\Windows\\SystemApps\\MicrosoftWindows.Client.CBS_cw5n1h2txyewy\\CrossDeviceResume.exe, [N/A]\n13756, C:\\Windows\\explorer.exe, [N/A]\n16592, C:\\Windows\\SystemApps\\Microsoft.Windows.StartMenuExperienceHost_cw5n1h2txyewy\\StartMenuExperienceHost.exe, [N/A]\n16496, C:\\Windows\\SystemApps\\MicrosoftWindows.Client.CBS_cw5n1h2txyewy\\SearchHost.exe, [N/A]\n20388, C:\\Windows\\SystemApps\\Microsoft.LockApp_cw5n1h2txyewy\\LockApp.exe, [N/A]\n19856, C:\\Program Files (x86)\\Microsoft\\EdgeWebView\\Application\\153.0.4234.48\\msedgewebview2.exe, [N/A]\n21864, C:\\Program Files\\NVIDIA Corporation\\NVIDIA App\\CEF\\NVIDIA Overlay.exe, [N/A]\n15368, C:\\Program Files\\NVIDIA Corporation\\NVIDIA App\\CEF\\NVIDIA Overlay.exe, [N/A]\n23204, C:\\Windows\\SystemApps\\MicrosoftWindows.Client.CBS_cw5n1h2txyewy\\TextInputHost.exe, [N/A]\n20596, C:\\Program Files\\LogiOptionsPlus\\logioptionsplus_agent.exe, [N/A]\n16776, C:\\Program Files\\WindowsApps\\Microsoft.YourPhone_1.26072.255.0_x64__8wekyb3d8bbwe\\PhoneExperienceHost.exe, [N/A]\n29352, C:\\Program Files\\WindowsApps\\Microsoft.6365217CE6EB4_102.2504.16004.0_x64__8wekyb3d8bbwe\\MicrosoftSecurityApp\\MicrosoftSecurityApp.exe, [N/A]\n5036, C:\\Program Files\\Intel\\Intel Graphics Software\\IntelGraphicsSoftware.exe, [N/A]\n12064, C:\\Program Files (x86)\\Microsoft\\EdgeWebView\\Application\\153.0.4234.48\\msedgewebview2.exe, [N/A]\n28020, C:\\Windows\\System32\\ApplicationFrameHost.exe, [N/A]\n30732, C:\\Windows\\ImmersiveControlPanel\\SystemSettings.exe, [N/A]\n30700, C:\\Users\\Admin\\AppData\\Local\\Programs\\MiniMark\\Runtime\\WebView2\\152.0.4191.53\\msedgewebview2.exe, [N/A]\n34240, C:\\Users\\Admin\\AppData\\Local\\Programs\\MiniMark\\MiniMark.Windows.exe, [N/A]\n19236, C:\\Windows\\SystemApps\\ShellExperienceHost_cw5n1h2txyewy\\ShellExperienceHost.exe, [N/A]\n3388, C:\\Program Files (x86)\\Microsoft\\Edge\\Application\\msedge.exe, [N/A]\n23864, C:\\Users\\Admin\\AppData\\Local\\Programs\\WorkBuddy\\WorkBuddy.exe, [N/A]\n30016, C:\\Program Files (x86)\\Microsoft\\Edge\\Application\\msedge.exe, [N/A]\n15804, C:\\Program Files\\WindowsApps\\Microsoft.WindowsTerminal_1.24.11911.0_x64__8wekyb3d8bbwe\\WindowsTerminal.exe, [N/A]\n28420, C:\\Windows\\System32\\ShellHost.exe, [N/A]\n4580, C:\\Python313\\python.exe, [N/A]\n",
+  "model_identity": {
+    "kind": "int4",
+    "base_revision": "b3179ad355be050328e483a9dfdd9e60cd62adfa",
+    "files": [
+      {
+        "path": "model_index.json",
+        "size": 447,
+        "sha256": "cf1ecd104ea090855d60d8cec0895c1e9d6ee41b2f87e231b678beecd1cf7809"
+      },
+      {
+        "path": "processor/added_tokens.json",
+        "size": 707,
+        "sha256": "c0284b582e14987fbd3d5a2cb2bd139084371ed9acbae488829a1c900833c680"
+      },
+      {
+        "path": "processor/chat_template.jinja",
+        "size": 5292,
+        "sha256": "3636d0f0bd6bef02654cdffdc447b79cb2cef8ab02cc75267345946291a489e4"
+      },
+      {
+        "path": "processor/merges.txt",
+        "size": 1671853,
+        "sha256": "8831e4f1a044471340f7c0a83d7bd71306a5b867e95fd870f74d0c5308a904d5"
+      },
+      {
+        "path": "processor/preprocessor_config.json",
+        "size": 782,
+        "sha256": "93585062a80db5e8ca038efc7726a3e6411d9db948472d81d63c6303993be8c5"
+      },
+      {
+        "path": "processor/special_tokens_map.json",
+        "size": 613,
+        "sha256": "76862e765266b85aa9459767e33cbaf13970f327a0e88d1c65846c2ddd3a1ecd"
+      },
+      {
+        "path": "processor/tokenizer.json",
+        "size": 11422654,
+        "sha256": "aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4"
+      },
+      {
+        "path": "processor/tokenizer_config.json",
+        "size": 5445,
+        "sha256": "81ec7bb9530159b326c0bef1d0b6c33d392090524014ea3f0123a3c1eb9c2af5"
+      },
+      {
+        "path": "processor/video_preprocessor_config.json",
+        "size": 817,
+        "sha256": "59c5c9eb52182eb14c06ffb10ca9effd29adce5f238a95de23ca14a38dbd2cb1"
+      },
+      {
+        "path": "processor/vocab.json",
+        "size": 2776833,
+        "sha256": "ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910"
+      },
+      {
+        "path": "scheduler/scheduler_config.json",
+        "size": 485,
+        "sha256": "5895f3a167c14a967fe9ac70c64924ae5acc79799e0679fd12907e594a713cd1"
+      },
+      {
+        "path": "text_encoder/config.json",
+        "size": 13320,
+        "sha256": "08f9c2ccf4763fa08769ebf0d5ff07e023b7849ab7fdb6be7f7f68c1c638dd15"
+      },
+      {
+        "path": "text_encoder/generation_config.json",
+        "size": 468,
+        "sha256": "013502676c64afb419718353b70249641b19f89fc8cc316177730bb957ecf9ed"
+      },
+      {
+        "path": "text_encoder/model-00001-of-00002.safetensors",
+        "size": 4986732760,
+        "sha256": "eb7c22e8c9b5597dbcee68d83f2e44f2eca29e5c942bbd22b4ff1743f293d30d"
+      },
+      {
+        "path": "text_encoder/model-00002-of-00002.safetensors",
+        "size": 2737783200,
+        "sha256": "e586e01002cc266f9794fdb7cfb1ea290a603f45a704ce08058ec15461532b4e"
+      },
+      {
+        "path": "text_encoder/model.safetensors.index.json",
+        "size": 167983,
+        "sha256": "d5c92794b905d777d3279918111a75b412805908fc1fdc773e6e681bce633cb6"
+      },
+      {
+        "path": "transformer/config.json",
+        "size": 5704,
+        "sha256": "62a3251c7145ee0d5ae271a324126b2f8548f69785fa1e3f0f4073d63704dece"
+      },
+      {
+        "path": "transformer/diffusion_pytorch_model.safetensors",
+        "size": 4365385520,
+        "sha256": "5c83319d84f2f985b289b5c69f0f98fd0a7b68a0b60b25550c36c24c33f32f5a"
+      },
+      {
+        "path": "vae/config.json",
+        "size": 2079,
+        "sha256": "9785d527b278cb8b210e9f48a8028d92a4190968ce7e24c8fc85d9d1b82f6ba2"
+      },
+      {
+        "path": "vae/diffusion_pytorch_model.safetensors",
+        "size": 1350989512,
+        "sha256": "a07a1b7c4ee2966a1b3bdc37de9b4f983d56937e46619f709a80b6e490675417"
+      }
+    ],
+    "fingerprint": "b80bd420b7a3a50aac8108adf32b2eca4b0258578f2eb4527524240376377fd4"
+  },
+  "model": "D:\\Git\\Qwen-Image-2.1-INT4\\models\\int4",
+  "load_seconds": 14.891308000005665,
+  "load_peak_allocated_bytes": 8704,
+  "offload": "model",
+  "requested_offload": "model",
+  "runtime_policy": {
+    "offload": "model",
+    "dtype": "bfloat16",
+    "use_stream": false,
+    "device": "cuda"
+  },
+  "warmup": true,
+  "after_load_memory": {
+    "allocated_bytes": 0,
+    "reserved_bytes": 0,
+    "cuda_free_bytes": 32494321664,
+    "cuda_total_bytes": 34162016256
+  },
+  "generator_device": "cpu",
+  "cases_sha256": "c0bc77b0aa28c9f11442c30cafd07d31bca72a179c7b759fb2a5ed9e328896c8"
+}

evaluation/int4/records.jsonl

@@ -0,0 +1,14 @@
+{"case_id": "portrait", "category": "portrait", "prompt": "A natural documentary portrait of an elderly woman with silver hair and freckles, wearing a blue wool sweater, standing beside a window in soft morning light. Realistic skin texture, gentle expression, waist-up composition.", "seed": 42, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 46137344, "cuda_free_bytes": 32299286528, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 79691776, "cuda_free_bytes": 32265732096, "cuda_total_bytes": 34162016256}, "seconds": 29.40627289999975, "peak_allocated_bytes": 8274727936, "peak_reserved_bytes": 9380560896, "process_rss_after_bytes": 18291404800, "image": "portrait-s42.png", "image_sha256": "a084dccd83c651b709a301925fcc085f005d2b4f49fac3ff0817d6d78644e7d3", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "portrait", "category": "portrait", "prompt": "A natural documentary portrait of an elderly woman with silver hair and freckles, wearing a blue wool sweater, standing beside a window in soft morning light. Realistic skin texture, gentle expression, waist-up composition.", "seed": 123, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 46137344, "cuda_free_bytes": 32299286528, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 79691776, "cuda_free_bytes": 32265732096, "cuda_total_bytes": 34162016256}, "seconds": 28.207035100000212, "peak_allocated_bytes": 8274727936, "peak_reserved_bytes": 9380560896, "process_rss_after_bytes": 18711670784, "image": "portrait-s123.png", "image_sha256": "0f3b5e3c9c2464d3f2567c8f6e9040494de1fdf06042fcb0696df58b8f58e047", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "english_text", "category": "typography_en", "prompt": "A clean editorial poster with a deep blue background. Large exact headline at the top: \"CREATE WITH LIGHT\". Smaller exact subtitle: \"September 2026\". A realistic orange desk lamp occupies the lower half. Elegant balanced typography, no additional text.", "seed": 42, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 46137344, "cuda_free_bytes": 32299286528, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 88080384, "cuda_free_bytes": 32257343488, "cuda_total_bytes": 34162016256}, "seconds": 28.350070699991193, "peak_allocated_bytes": 8279566336, "peak_reserved_bytes": 9386852352, "process_rss_after_bytes": 18880417792, "image": "english_text-s42.png", "image_sha256": "d6880ef09d3d5986da086e012b89e4a5fabfde16e16721e49c0436baadec53da", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "english_text", "category": "typography_en", "prompt": "A clean editorial poster with a deep blue background. Large exact headline at the top: \"CREATE WITH LIGHT\". Smaller exact subtitle: \"September 2026\". A realistic orange desk lamp occupies the lower half. Elegant balanced typography, no additional text.", "seed": 123, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 46137344, "cuda_free_bytes": 32299286528, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 88080384, "cuda_free_bytes": 32257343488, "cuda_total_bytes": 34162016256}, "seconds": 28.674008000001777, "peak_allocated_bytes": 8279566336, "peak_reserved_bytes": 9386852352, "process_rss_after_bytes": 18889183232, "image": "english_text-s123.png", "image_sha256": "add2118fcebc3e2f63788382b323813fe7d967f1e88c91ae271fec057156e3a1", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "chinese_text", "category": "typography_zh", "prompt": "设计一张现代咖啡馆海报，米白背景，中央一杯拉花拿铁。顶部准确写上中文大标题“慢下来，喝杯咖啡”，下方准确写上“小店今日营业”。文字清晰，留白充分，暖色摄影，不添加其他文字。", "seed": 42, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 46137344, "cuda_free_bytes": 32299286528, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 102760448, "cuda_free_bytes": 32242663424, "cuda_total_bytes": 34162016256}, "seconds": 28.668083800002933, "peak_allocated_bytes": 8281985024, "peak_reserved_bytes": 9097445376, "process_rss_after_bytes": 19008032768, "image": "chinese_text-s42.png", "image_sha256": "61387812385423c8c240d7f34875fd3b2b88cc1662db3ce084757c434915d1e8", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "chinese_text", "category": "typography_zh", "prompt": "设计一张现代咖啡馆海报，米白背景，中央一杯拉花拿铁。顶部准确写上中文大标题“慢下来，喝杯咖啡”，下方准确写上“小店今日营业”。文字清晰，留白充分，暖色摄影，不添加其他文字。", "seed": 123, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 46137344, "cuda_free_bytes": 32299286528, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 100663296, "cuda_free_bytes": 32244760576, "cuda_total_bytes": 34162016256}, "seconds": 28.79211700000451, "peak_allocated_bytes": 8281985024, "peak_reserved_bytes": 9395240960, "process_rss_after_bytes": 19151949824, "image": "chinese_text-s123.png", "image_sha256": "b1740259fff3ebc07d14c12a837a899267a468ff4318a53c163815a3d6be53fb", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "composition", "category": "spatial_counting", "prompt": "A studio photograph on a light gray tabletop: exactly three ceramic cups in a row, a red cup on the left, a blue cup in the middle, and a yellow cup on the right. A single green apple sits in front of the blue cup. Soft shadows, no text.", "seed": 42, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 46137344, "cuda_free_bytes": 32299286528, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 98566144, "cuda_free_bytes": 32246857728, "cuda_total_bytes": 34162016256}, "seconds": 28.723114799999166, "peak_allocated_bytes": 8281501696, "peak_reserved_bytes": 9397338112, "process_rss_after_bytes": 19240718336, "image": "composition-s42.png", "image_sha256": "2074e38fd2af7d7cc8cf2f483553fdfbeeaad65094e70062462c0fa9b48b6f29", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "composition", "category": "spatial_counting", "prompt": "A studio photograph on a light gray tabletop: exactly three ceramic cups in a row, a red cup on the left, a blue cup in the middle, and a yellow cup on the right. A single green apple sits in front of the blue cup. Soft shadows, no text.", "seed": 123, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 46137344, "cuda_free_bytes": 32299286528, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 98566144, "cuda_free_bytes": 32246857728, "cuda_total_bytes": 34162016256}, "seconds": 28.66992489999393, "peak_allocated_bytes": 8281501696, "peak_reserved_bytes": 9397338112, "process_rss_after_bytes": 19314692096, "image": "composition-s123.png", "image_sha256": "c7b45122d69e8b4720ab9e3877f86a8da172629ccc0335ce303195b5000944e8", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "texture", "category": "texture", "prompt": "Macro photography of a small kingfisher perched on a mossy branch beside clear water, detailed blue feathers, droplets, natural sunlight, softly blurred forest background, realistic textures.", "seed": 42, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 46137344, "cuda_free_bytes": 32299286528, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 79691776, "cuda_free_bytes": 32265732096, "cuda_total_bytes": 34162016256}, "seconds": 28.9771555000043, "peak_allocated_bytes": 8271822848, "peak_reserved_bytes": 9378463744, "process_rss_after_bytes": 19233792000, "image": "texture-s42.png", "image_sha256": "2f4809ffb0301e2ddddf66ddfb834d1e750175a675fb05067cb671292ff50187", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "texture", "category": "texture", "prompt": "Macro photography of a small kingfisher perched on a mossy branch beside clear water, detailed blue feathers, droplets, natural sunlight, softly blurred forest background, realistic textures.", "seed": 123, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 48234496, "cuda_free_bytes": 32297189376, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 81788928, "cuda_free_bytes": 32263634944, "cuda_total_bytes": 34162016256}, "seconds": 28.759817099999054, "peak_allocated_bytes": 8271822848, "peak_reserved_bytes": 9378463744, "process_rss_after_bytes": 19198152704, "image": "texture-s123.png", "image_sha256": "7c50f70eb4f31c8f4322330b40a2cfd892640157abcc2f4ded826b6f7b8345f6", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "rgba", "category": "transparency", "prompt": "This is an RGBA image with transparency. A cute cartoon dragon sticker, full body, green scales and small orange wings. The image has alpha channel and the background is transparent.", "seed": 42, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 48234496, "cuda_free_bytes": 32297189376, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 81788928, "cuda_free_bytes": 32263634944, "cuda_total_bytes": 34162016256}, "seconds": 28.55306270001165, "peak_allocated_bytes": 8270855168, "peak_reserved_bytes": 9378463744, "process_rss_after_bytes": 19225985024, "image": "rgba-s42.png", "image_sha256": "47635b9cf96c9d47991df7c39b2bf8a5bd7c4b8e05ddea33034b83eff7e10f7c", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "rgba", "category": "transparency", "prompt": "This is an RGBA image with transparency. A cute cartoon dragon sticker, full body, green scales and small orange wings. The image has alpha channel and the background is transparent.", "seed": 123, "source_seed": null, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": null, "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 48234496, "cuda_free_bytes": 32297189376, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 13239296, "reserved_bytes": 81788928, "cuda_free_bytes": 32263634944, "cuda_total_bytes": 34162016256}, "seconds": 28.541914399989764, "peak_allocated_bytes": 8270855168, "peak_reserved_bytes": 9378463744, "process_rss_after_bytes": 19322126336, "image": "rgba-s123.png", "image_sha256": "fb13ec3ed4d00651e1050c4b1c0f08ef47cd85f179e3b3bf516d58dde6dcd2ef", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "edit", "category": "image_editing", "prompt": "Change only the blue sweater to a red sweater. Preserve the same person, face, pose, lighting and background.", "seed": 1000042, "source_seed": 42, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": "01e37e178fb0d35d5c706b7edd76bc1a6c5b50ec3536901a467fd45295d097aa", "before_memory": {"allocated_bytes": 13239296, "reserved_bytes": 48234496, "cuda_free_bytes": 32297189376, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 14288384, "reserved_bytes": 3055550464, "cuda_free_bytes": 29289873408, "cuda_total_bytes": 34162016256}, "seconds": 33.80973500000255, "peak_allocated_bytes": 10722032640, "peak_reserved_bytes": 11874074624, "process_rss_after_bytes": 19633192960, "image": "edit-s1000042.png", "image_sha256": "ae986b19def44d5bb3fa42e0338322a5e15177dfd7b909c4241c0c299f7c0f7e", "image_mode": "RGBA", "actual_size": [1024, 1024]}
+{"case_id": "edit", "category": "image_editing", "prompt": "Change only the blue sweater to a red sweater. Preserve the same person, face, pose, lighting and background.", "seed": 1000123, "source_seed": 42, "width": 1024, "height": 1024, "steps": 40, "cfg": 1.0, "kv_cache": true, "offload": "model", "input_sha256": "01e37e178fb0d35d5c706b7edd76bc1a6c5b50ec3536901a467fd45295d097aa", "before_memory": {"allocated_bytes": 14288384, "reserved_bytes": 50331648, "cuda_free_bytes": 32295092224, "cuda_total_bytes": 34162016256}, "after_memory": {"allocated_bytes": 14288384, "reserved_bytes": 3055550464, "cuda_free_bytes": 29289873408, "cuda_total_bytes": 34162016256}, "seconds": 33.51434729999164, "peak_allocated_bytes": 10722032640, "peak_reserved_bytes": 11574181888, "process_rss_after_bytes": 19715796992, "image": "edit-s1000123.png", "image_sha256": "c60c71631e6e0522d4ef29b13be4354231a09a14d3768881e9cfee328c6dc491", "image_mode": "RGBA", "actual_size": [1024, 1024]}

evaluation/qualitative.md

@@ -0,0 +1,39 @@
+# Visual inspection
+
+These notes describe images I looked at. They are not a score, and they do not cover every seed. Pixel drift for all 14 pairs is in comparison.csv.
+
+portrait, seed 42: Both are waist-up photographs of an older woman in a blue wool sweater beside a window. The face, the hair, and how much of the window is in frame differ. Skin and knit texture stay photographic in both.
+
+english_text, seed 42: Both render the headline CREATE WITH LIGHT and the subtitle September 2026. The orange lamp, books, and notebook are arranged differently. I did not see an extra headline.
+
+chinese_text, seed 42: Both render 慢下来，喝杯咖啡 and 小店今日营业 above a latte. Both also add a footer of unwanted, partly illegible text, so neither follows “不添加其他文字” completely.
+
+composition, seed 42: Both show exactly three cups, red then blue then yellow, and one green apple in front of the blue cup. Cup shape and spacing differ. The INT4 cups have handles; the BF16 cups do not.
+
+texture, seed 42: Both are plausible kingfishers on a mossy branch over water. The INT4 bird faces left and the BF16 bird faces right. Feather and bark detail are visible in both.
+
+rgba, seed 42: Both are full-body green cartoon dragons with orange wings, saved as RGBA. Measured alpha runs from 0 to 255. Pixels with alpha at or below 5 are 63.5% of the BF16 image and 73.0% of the INT4 image. The BF16 dragon has a white sticker outline.
+
+edit, seed 1000042: The input is the BF16 portrait from seed 42, and the edit seed is not 42. Both change the sweater from blue to red and keep the same face, window light, and pose. This pair is not visibly oversharpened. That does not show that every edit will preserve identity.
+
+The 8GB-cap portrait (group offload, CPU VAE encode and decode) is a separate run from this paired set. On that image the woman, blue sweater, and window are all present. The same cap also completed both edit seeds, 1000042 and 1000123, at 1024 for 40 steps. Allocated peaks were 3.48 GiB and reserved peaks were 4.08 and 4.24 GiB.
+
+# 目视记录
+
+下面只记录看过的图，不是评分，也没有逐张覆盖第二个种子。14 对的像素差在 comparison.csv。
+
+portrait，种子 42：两种精度都是窗边、蓝毛衣、半身的老年女性照片。脸、头发和窗户入画的多少不同。皮肤和针织纹理都还像照片。
+
+english_text，种子 42：两种精度都写出了 CREATE WITH LIGHT 和 September 2026。橙色台灯、书和笔记本的摆放不同。没有看到多出来的主标题。
+
+chinese_text，种子 42：两种精度都写出了“慢下来，喝杯咖啡”和“小店今日营业”，中间是拿铁。两种精度的页脚都出现了多余且部分无法辨认的文字，都没有完全遵守“不添加其他文字”。
+
+composition，种子 42：两种精度都是左红、中蓝、右黄三个杯子，蓝杯前有一个青苹果。杯子形状和间距不同。INT4 的杯子有把手，BF16 没有。
+
+texture，种子 42：两种精度都是水边苔枝上的翠鸟。INT4 的鸟朝左，BF16 的鸟朝右。羽毛和树皮细节都还在。
+
+rgba，种子 42：两种精度都是全身、绿鳞、橙翼的卡通龙，文件为 RGBA。alpha 范围是 0 到 255。alpha 小于等于 5 的像素，BF16 占 63.5%，INT4 占 73.0%。BF16 的龙有一圈白色贴纸描边。
+
+edit，种子 1000042：输入是 BF16 种子 42 的肖像，编辑种子不是 42。两种精度都把蓝毛衣改成了红毛衣，脸、窗光和姿态还在。这一对没有明显过锐。这不能说明每次编辑都会保住身份。
+
+8GB 分配上限下的肖像是另一次运行（分组卸载，VAE 的编码和解码都在 CPU）。那张图里仍有人物、蓝毛衣和窗户。同一次上限设置也完成了两个编辑种子 1000042 和 1000123，1024、40 步。已分配峰值都是 3.48 GiB，保留峰值分别是 4.08 GiB 和 4.24 GiB。

evaluation/report.md

@@ -0,0 +1,33 @@
+# Informal release evaluation: BF16 / INT4
+
+This community comparison is reference material. It is not an official evaluation.
+
+14 paired outputs on NVIDIA GeForce RTX 5090; 1024×1024, 40 steps, offload=model, CFG=1, KV cache enabled.
+
+Warmup excluded. Pixel metrics measure drift. They are not a semantic quality score. The suite is small and does not establish a ranking.
+
+| Case | Seed | BF16 s | INT4 s | BF16 peak GiB | INT4 peak GiB | RGB MAE |
+|---|---:|---:|---:|---:|---:|---:|
+| portrait | 42 | 28.05 | 29.41 | 16.40 | 7.71 | 0.0528 |
+| portrait | 123 | 26.06 | 28.21 | 16.40 | 7.71 | 0.0639 |
+| english_text | 42 | 25.40 | 28.35 | 16.41 | 7.71 | 0.0839 |
+| english_text | 123 | 25.23 | 28.67 | 16.41 | 7.71 | 0.0547 |
+| chinese_text | 42 | 25.55 | 28.67 | 16.41 | 7.71 | 0.0791 |
+| chinese_text | 123 | 25.61 | 28.79 | 16.41 | 7.71 | 0.0819 |
+| composition | 42 | 25.21 | 28.72 | 16.41 | 7.71 | 0.0297 |
+| composition | 123 | 25.51 | 28.67 | 16.41 | 7.71 | 0.0594 |
+| texture | 42 | 25.40 | 28.98 | 16.40 | 7.70 | 0.0775 |
+| texture | 123 | 25.13 | 28.76 | 16.40 | 7.70 | 0.0489 |
+| rgba | 42 | 25.02 | 28.55 | 16.40 | 7.70 | 0.0444 |
+| rgba | 123 | 25.76 | 28.54 | 16.40 | 7.70 | 0.0760 |
+| edit | 1000042 | 38.67 | 33.81 | 19.08 | 9.99 | 0.0112 |
+| edit | 1000123 | 31.10 | 33.51 | 19.08 | 9.99 | 0.0107 |
+
+## Summary
+
+Mean latency: BF16 26.98s; INT4 29.40s.
+Maximum allocated CUDA memory: BF16 19.08 GiB; INT4 9.99 GiB.
+
+Raw records: comparison.csv, bf16/records.jsonl, int4/records.jsonl.
+Visual notes belong in qualitative.md and are written after inspecting the images.
+

evaluation/summary.json

@@ -0,0 +1,14 @@
+{
+  "pairs": 14,
+  "cases": 7,
+  "bf16_mean_seconds": 26.97865971428629,
+  "int4_mean_seconds": 29.403332799999458,
+  "bf16_max_allocated_gib": 19.081242561340332,
+  "int4_max_allocated_gib": 9.985671043395996,
+  "width": 1024,
+  "height": 1024,
+  "steps": 40,
+  "warmup": true,
+  "bf16_weight_bytes": 33115613408,
+  "int4_weight_bytes": 13440890992
+}

evaluation/vram8/environment.json

@@ -0,0 +1,165 @@
+{
+  "python": "3.13.5",
+  "platform": "Windows-11-10.0.26200-SP0",
+  "gpu": "NVIDIA GeForce RTX 5090",
+  "cuda": "12.8",
+  "mps": false,
+  "system_ram_bytes": 274298757120,
+  "packages": {
+    "torch": "2.10.0+cu128",
+    "diffusers": "0.41.0.dev0",
+    "transformers": "5.17.0",
+    "accelerate": "1.15.0",
+    "sdnq": "0.2.7",
+    "safetensors": "0.8.0",
+    "huggingface-hub": "1.32.0",
+    "tokenizers": "0.23.2",
+    "numpy": "2.2.6",
+    "pillow": "12.2.0",
+    "psutil": "7.0.0"
+  },
+  "pid": 32432,
+  "before_load_memory": {
+    "allocated_bytes": 0,
+    "reserved_bytes": 0,
+    "cuda_free_bytes": 32496418816,
+    "cuda_total_bytes": 34162016256
+  },
+  "memory_cap": {
+    "cap_bytes": 7730941132,
+    "cap_gib": 7.2,
+    "total_bytes": 34162016256
+  },
+  "benchmark_sha256": "83bdbf97778874d8d7b2a605e357940cb6a5e0f3e05bed12bd3f2d176deceb06",
+  "runtime_helper_sha256": "81ea21948d27f61ba210b8fc44487b057200b741df0ab8c587f60d9cec02f6f9",
+  "device_helper_sha256": "aab115693753a43732b52873e11ec2704855f8631b52ddacdd7fc976f55da3cd",
+  "gpu_before": "name, driver_version, memory.used [MiB], memory.free [MiB]\nNVIDIA GeForce RTX 5090, 616.92, 2998 MiB, 29190 MiB\n",
+  "gpu_processes_before": "pid, process_name, used_gpu_memory [MiB]\n14108, C:\\Windows\\SystemApps\\MicrosoftWindows.Client.CBS_cw5n1h2txyewy\\CrossDeviceResume.exe, [N/A]\n13756, C:\\Windows\\explorer.exe, [N/A]\n16592, C:\\Windows\\SystemApps\\Microsoft.Windows.StartMenuExperienceHost_cw5n1h2txyewy\\StartMenuExperienceHost.exe, [N/A]\n16496, C:\\Windows\\SystemApps\\MicrosoftWindows.Client.CBS_cw5n1h2txyewy\\SearchHost.exe, [N/A]\n20388, C:\\Windows\\SystemApps\\Microsoft.LockApp_cw5n1h2txyewy\\LockApp.exe, [N/A]\n19856, C:\\Program Files (x86)\\Microsoft\\EdgeWebView\\Application\\153.0.4234.48\\msedgewebview2.exe, [N/A]\n21864, C:\\Program Files\\NVIDIA Corporation\\NVIDIA App\\CEF\\NVIDIA Overlay.exe, [N/A]\n15368, C:\\Program Files\\NVIDIA Corporation\\NVIDIA App\\CEF\\NVIDIA Overlay.exe, [N/A]\n23204, C:\\Windows\\SystemApps\\MicrosoftWindows.Client.CBS_cw5n1h2txyewy\\TextInputHost.exe, [N/A]\n20596, C:\\Program Files\\LogiOptionsPlus\\logioptionsplus_agent.exe, [N/A]\n16776, C:\\Program Files\\WindowsApps\\Microsoft.YourPhone_1.26072.255.0_x64__8wekyb3d8bbwe\\PhoneExperienceHost.exe, [N/A]\n29352, C:\\Program Files\\WindowsApps\\Microsoft.6365217CE6EB4_102.2504.16004.0_x64__8wekyb3d8bbwe\\MicrosoftSecurityApp\\MicrosoftSecurityApp.exe, [N/A]\n5036, C:\\Program Files\\Intel\\Intel Graphics Software\\IntelGraphicsSoftware.exe, [N/A]\n12064, C:\\Program Files (x86)\\Microsoft\\EdgeWebView\\Application\\153.0.4234.48\\msedgewebview2.exe, [N/A]\n28020, C:\\Windows\\System32\\ApplicationFrameHost.exe, [N/A]\n30732, C:\\Windows\\ImmersiveControlPanel\\SystemSettings.exe, [N/A]\n30700, C:\\Users\\Admin\\AppData\\Local\\Programs\\MiniMark\\Runtime\\WebView2\\152.0.4191.53\\msedgewebview2.exe, [N/A]\n34240, C:\\Users\\Admin\\AppData\\Local\\Programs\\MiniMark\\MiniMark.Windows.exe, [N/A]\n19236, C:\\Windows\\SystemApps\\ShellExperienceHost_cw5n1h2txyewy\\ShellExperienceHost.exe, [N/A]\n3388, C:\\Program Files (x86)\\Microsoft\\Edge\\Application\\msedge.exe, [N/A]\n23864, C:\\Users\\Admin\\AppData\\Local\\Programs\\WorkBuddy\\WorkBuddy.exe, [N/A]\n30016, C:\\Program Files (x86)\\Microsoft\\Edge\\Application\\msedge.exe, [N/A]\n15804, C:\\Program Files\\WindowsApps\\Microsoft.WindowsTerminal_1.24.11911.0_x64__8wekyb3d8bbwe\\WindowsTerminal.exe, [N/A]\n28420, C:\\Windows\\System32\\ShellHost.exe, [N/A]\n32432, C:\\Python313\\python.exe, [N/A]\n",
+  "model_identity": {
+    "kind": "int4",
+    "base_revision": "b3179ad355be050328e483a9dfdd9e60cd62adfa",
+    "files": [
+      {
+        "path": "model_index.json",
+        "size": 447,
+        "sha256": "cf1ecd104ea090855d60d8cec0895c1e9d6ee41b2f87e231b678beecd1cf7809"
+      },
+      {
+        "path": "processor/added_tokens.json",
+        "size": 707,
+        "sha256": "c0284b582e14987fbd3d5a2cb2bd139084371ed9acbae488829a1c900833c680"
+      },
+      {
+        "path": "processor/chat_template.jinja",
+        "size": 5292,
+        "sha256": "3636d0f0bd6bef02654cdffdc447b79cb2cef8ab02cc75267345946291a489e4"
+      },
+      {
+        "path": "processor/merges.txt",
+        "size": 1671853,
+        "sha256": "8831e4f1a044471340f7c0a83d7bd71306a5b867e95fd870f74d0c5308a904d5"
+      },
+      {
+        "path": "processor/preprocessor_config.json",
+        "size": 782,
+        "sha256": "93585062a80db5e8ca038efc7726a3e6411d9db948472d81d63c6303993be8c5"
+      },
+      {
+        "path": "processor/special_tokens_map.json",
+        "size": 613,
+        "sha256": "76862e765266b85aa9459767e33cbaf13970f327a0e88d1c65846c2ddd3a1ecd"
+      },
+      {
+        "path": "processor/tokenizer.json",
+        "size": 11422654,
+        "sha256": "aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4"
+      },
+      {
+        "path": "processor/tokenizer_config.json",
+        "size": 5445,
+        "sha256": "81ec7bb9530159b326c0bef1d0b6c33d392090524014ea3f0123a3c1eb9c2af5"
+      },
+      {
+        "path": "processor/video_preprocessor_config.json",
+        "size": 817,
+        "sha256": "59c5c9eb52182eb14c06ffb10ca9effd29adce5f238a95de23ca14a38dbd2cb1"
+      },
+      {
+        "path": "processor/vocab.json",
+        "size": 2776833,
+        "sha256": "ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910"
+      },
+      {
+        "path": "scheduler/scheduler_config.json",
+        "size": 485,
+        "sha256": "5895f3a167c14a967fe9ac70c64924ae5acc79799e0679fd12907e594a713cd1"
+      },
+      {
+        "path": "text_encoder/config.json",
+        "size": 13320,
+        "sha256": "08f9c2ccf4763fa08769ebf0d5ff07e023b7849ab7fdb6be7f7f68c1c638dd15"
+      },
+      {
+        "path": "text_encoder/generation_config.json",
+        "size": 468,
+        "sha256": "013502676c64afb419718353b70249641b19f89fc8cc316177730bb957ecf9ed"
+      },
+      {
+        "path": "text_encoder/model-00001-of-00002.safetensors",
+        "size": 4986732760,
+        "sha256": "eb7c22e8c9b5597dbcee68d83f2e44f2eca29e5c942bbd22b4ff1743f293d30d"
+      },
+      {
+        "path": "text_encoder/model-00002-of-00002.safetensors",
+        "size": 2737783200,
+        "sha256": "e586e01002cc266f9794fdb7cfb1ea290a603f45a704ce08058ec15461532b4e"
+      },
+      {
+        "path": "text_encoder/model.safetensors.index.json",
+        "size": 167983,
+        "sha256": "d5c92794b905d777d3279918111a75b412805908fc1fdc773e6e681bce633cb6"
+      },
+      {
+        "path": "transformer/config.json",
+        "size": 5704,
+        "sha256": "62a3251c7145ee0d5ae271a324126b2f8548f69785fa1e3f0f4073d63704dece"
+      },
+      {
+        "path": "transformer/diffusion_pytorch_model.safetensors",
+        "size": 4365385520,
+        "sha256": "5c83319d84f2f985b289b5c69f0f98fd0a7b68a0b60b25550c36c24c33f32f5a"
+      },
+      {
+        "path": "vae/config.json",
+        "size": 2079,
+        "sha256": "9785d527b278cb8b210e9f48a8028d92a4190968ce7e24c8fc85d9d1b82f6ba2"
+      },
+      {
+        "path": "vae/diffusion_pytorch_model.safetensors",
+        "size": 1350989512,
+        "sha256": "a07a1b7c4ee2966a1b3bdc37de9b4f983d56937e46619f709a80b6e490675417"
+      }
+    ],
+    "fingerprint": "b80bd420b7a3a50aac8108adf32b2eca4b0258578f2eb4527524240376377fd4"
+  },
+  "model": "D:\\Git\\Qwen-Image-2.1-INT4\\models\\int4",
+  "load_seconds": 14.512587799996254,
+  "load_peak_allocated_bytes": 8704,
+  "offload": "group",
+  "requested_offload": "group",
+  "runtime_policy": {
+    "offload": "group",
+    "dtype": "bfloat16",
+    "use_stream": false,
+    "device": "cuda"
+  },
+  "warmup": true,
+  "after_load_memory": {
+    "allocated_bytes": 0,
+    "reserved_bytes": 0,
+    "cuda_free_bytes": 32494321664,
+    "cuda_total_bytes": 34162016256
+  },
+  "generator_device": "cpu",
+  "cases_sha256": "c0bc77b0aa28c9f11442c30cafd07d31bca72a179c7b759fb2a5ed9e328896c8"
+}