diff --git a/.gitattributes b/.gitattributes index a6344aac8c09253b3b630fb776ae94478aa0275b..111230162776a22227e04824830a1742cbd34bbf 100644 --- a/.gitattributes +++ b/.gitattributes @@ -33,3 +33,27 @@ 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 +QWENVL-PRIVATE/cookbooks/assets/agent_function_call/mobile_en_example.png filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/agent_function_call/mobile_zh_example.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/computer_use/computer_use1.jpeg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/computer_use/computer_use2.jpeg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/document_parsing/docparsing_example1.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/document_parsing/docparsing_example2.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/document_parsing/docparsing_example3.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/document_parsing/docparsing_example4.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/document_parsing/docparsing_example5.png filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/document_parsing/docparsing_example6.png filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/document_parsing/docparsing_example7.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/document_parsing/docparsing_example8.png filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/ocr/ocr_example1.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/ocr/ocr_example3.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/ocr/ocr_example5.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/spatial_understanding/Origamis.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/spatial_understanding/cakes.png filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/universal_recognition/unireco_bird_example.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/universal_recognition/unireco_birds_example.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/universal_recognition/unireco_celebrities_example.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/cookbooks/assets/universal_recognition/unireco_landmarks_example.jpg filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/qwen-vl-finetune/demo/videos/v_7bUu05RIksU.mp4 filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/qwen-vl-finetune/demo/videos/v_TpB_zMG3XBA.mp4 filter=lfs diff=lfs merge=lfs -text +QWENVL-PRIVATE/qwen-vl-finetune/demo/videos/v_rBMQFpHspmo.mp4 filter=lfs diff=lfs merge=lfs -text diff --git a/QWENVL-PRIVATE/.gitignore b/QWENVL-PRIVATE/.gitignore new file mode 100644 index 0000000000000000000000000000000000000000..1d9318c0b4956aa48b05832ab676071ed73ccc9f --- /dev/null +++ b/QWENVL-PRIVATE/.gitignore @@ -0,0 +1,17 @@ +# python generated files +__pycache__/ +*.py[oc] +build/ +dist/ +wheels/ +*.egg-info + +# venv +.venv +/exclude/ +.tmp.driveupload +/excluded/ +/output/ +.idea/ +logs/ +lmms-eval-0.5.0/ diff --git a/QWENVL-PRIVATE/LICENSE b/QWENVL-PRIVATE/LICENSE new file mode 100644 index 0000000000000000000000000000000000000000..973f6ce469bf97dadfff70a3a262cb9e7dce40a1 --- /dev/null +++ b/QWENVL-PRIVATE/LICENSE @@ -0,0 +1,352 @@ +Creative Commons Attribution-NonCommercial 4.0 International + +Creative Commons Corporation ("Creative Commons") is not a law firm and +does not provide legal services or legal advice. Distribution of +Creative Commons public licenses does not create a lawyer-client or +other relationship. Creative Commons makes its licenses and related +information available on an "as-is" basis. Creative Commons gives no +warranties regarding its licenses, any material licensed under their +terms and conditions, or any related information. Creative Commons +disclaims all liability for damages resulting from their use to the +fullest extent possible. + +Using Creative Commons Public Licenses + +Creative Commons public licenses provide a standard set of terms and +conditions that creators and other rights holders may use to share +original works of authorship and other material subject to copyright and +certain other rights specified in the public license below. The +following considerations are for informational purposes only, are not +exhaustive, and do not form part of our licenses. + +- Considerations for licensors: Our public licenses are intended for + use by those authorized to give the public permission to use + material in ways otherwise restricted by copyright and certain other + rights. Our licenses are irrevocable. Licensors should read and + understand the terms and conditions of the license they choose + before applying it. Licensors should also secure all rights + necessary before applying our licenses so that the public can reuse + the material as expected. Licensors should clearly mark any material + not subject to the license. This includes other CC-licensed + material, or material used under an exception or limitation to + copyright. More considerations for licensors : + wiki.creativecommons.org/Considerations_for_licensors + +- Considerations for the public: By using one of our public licenses, + a licensor grants the public permission to use the licensed material + under specified terms and conditions. If the licensor's permission + is not necessary for any reason–for example, because of any + applicable exception or limitation to copyright–then that use is not + regulated by the license. Our licenses grant only permissions under + copyright and certain other rights that a licensor has authority to + grant. Use of the licensed material may still be restricted for + other reasons, including because others have copyright or other + rights in the material. A licensor may make special requests, such + as asking that all changes be marked or described. Although not + required by our licenses, you are encouraged to respect those + requests where reasonable. More considerations for the public : + wiki.creativecommons.org/Considerations_for_licensees + +Creative Commons Attribution-NonCommercial 4.0 International Public +License + +By exercising the Licensed Rights (defined below), You accept and agree +to be bound by the terms and conditions of this Creative Commons +Attribution-NonCommercial 4.0 International Public License ("Public +License"). To the extent this Public License may be interpreted as a +contract, You are granted the Licensed Rights in consideration of Your +acceptance of these terms and conditions, and the Licensor grants You +such rights in consideration of benefits the Licensor receives from +making the Licensed Material available under these terms and conditions. + +- Section 1 – Definitions. + + - a. Adapted Material means material subject to Copyright and + Similar Rights that is derived from or based upon the Licensed + Material and in which the Licensed Material is translated, + altered, arranged, transformed, or otherwise modified in a + manner requiring permission under the Copyright and Similar + Rights held by the Licensor. For purposes of this Public + License, where the Licensed Material is a musical work, + performance, or sound recording, Adapted Material is always + produced where the Licensed Material is synched in timed + relation with a moving image. + - b. Adapter's License means the license You apply to Your + Copyright and Similar Rights in Your contributions to Adapted + Material in accordance with the terms and conditions of this + Public License. + - c. Copyright and Similar Rights means copyright and/or similar + rights closely related to copyright including, without + limitation, performance, broadcast, sound recording, and Sui + Generis Database Rights, without regard to how the rights are + labeled or categorized. For purposes of this Public License, the + rights specified in Section 2(b)(1)-(2) are not Copyright and + Similar Rights. + - d. Effective Technological Measures means those measures that, + in the absence of proper authority, may not be circumvented + under laws fulfilling obligations under Article 11 of the WIPO + Copyright Treaty adopted on December 20, 1996, and/or similar + international agreements. + - e. Exceptions and Limitations means fair use, fair dealing, + and/or any other exception or limitation to Copyright and + Similar Rights that applies to Your use of the Licensed + Material. + - f. Licensed Material means the artistic or literary work, + database, or other material to which the Licensor applied this + Public License. + - g. Licensed Rights means the rights granted to You subject to + the terms and conditions of this Public License, which are + limited to all Copyright and Similar Rights that apply to Your + use of the Licensed Material and that the Licensor has authority + to license. + - h. Licensor means the individual(s) or entity(ies) granting + rights under this Public License. + - i. NonCommercial means not primarily intended for or directed + towards commercial advantage or monetary compensation. For + purposes of this Public License, the exchange of the Licensed + Material for other material subject to Copyright and Similar + Rights by digital file-sharing or similar means is NonCommercial + provided there is no payment of monetary compensation in + connection with the exchange. + - j. Share means to provide material to the public by any means or + process that requires permission under the Licensed Rights, such + as reproduction, public display, public performance, + distribution, dissemination, communication, or importation, and + to make material available to the public including in ways that + members of the public may access the material from a place and + at a time individually chosen by them. + - k. Sui Generis Database Rights means rights other than copyright + resulting from Directive 96/9/EC of the European Parliament and + of the Council of 11 March 1996 on the legal protection of + databases, as amended and/or succeeded, as well as other + essentially equivalent rights anywhere in the world. + - l. You means the individual or entity exercising the Licensed + Rights under this Public License. Your has a corresponding + meaning. + +- Section 2 – Scope. + + - a. License grant. + - 1. Subject to the terms and conditions of this Public + License, the Licensor hereby grants You a worldwide, + royalty-free, non-sublicensable, non-exclusive, irrevocable + license to exercise the Licensed Rights in the Licensed + Material to: + - A. reproduce and Share the Licensed Material, in whole + or in part, for NonCommercial purposes only; and + - B. produce, reproduce, and Share Adapted Material for + NonCommercial purposes only. + - 2. Exceptions and Limitations. For the avoidance of doubt, + where Exceptions and Limitations apply to Your use, this + Public License does not apply, and You do not need to comply + with its terms and conditions. + - 3. Term. The term of this Public License is specified in + Section 6(a). + - 4. Media and formats; technical modifications allowed. The + Licensor authorizes You to exercise the Licensed Rights in + all media and formats whether now known or hereafter + created, and to make technical modifications necessary to do + so. The Licensor waives and/or agrees not to assert any + right or authority to forbid You from making technical + modifications necessary to exercise the Licensed Rights, + including technical modifications necessary to circumvent + Effective Technological Measures. For purposes of this + Public License, simply making modifications authorized by + this Section 2(a)(4) never produces Adapted Material. + - 5. Downstream recipients. + - A. Offer from the Licensor – Licensed Material. Every + recipient of the Licensed Material automatically + receives an offer from the Licensor to exercise the + Licensed Rights under the terms and conditions of this + Public License. + - B. No downstream restrictions. You may not offer or + impose any additional or different terms or conditions + on, or apply any Effective Technological Measures to, + the Licensed Material if doing so restricts exercise of + the Licensed Rights by any recipient of the Licensed + Material. + - 6. No endorsement. Nothing in this Public License + constitutes or may be construed as permission to assert or + imply that You are, or that Your use of the Licensed + Material is, connected with, or sponsored, endorsed, or + granted official status by, the Licensor or others + designated to receive attribution as provided in Section + 3(a)(1)(A)(i). + - b. Other rights. + - 1. Moral rights, such as the right of integrity, are not + licensed under this Public License, nor are publicity, + privacy, and/or other similar personality rights; however, + to the extent possible, the Licensor waives and/or agrees + not to assert any such rights held by the Licensor to the + limited extent necessary to allow You to exercise the + Licensed Rights, but not otherwise. + - 2. Patent and trademark rights are not licensed under this + Public License. + - 3. To the extent possible, the Licensor waives any right to + collect royalties from You for the exercise of the Licensed + Rights, whether directly or through a collecting society + under any voluntary or waivable statutory or compulsory + licensing scheme. In all other cases the Licensor expressly + reserves any right to collect such royalties, including when + the Licensed Material is used other than for NonCommercial + purposes. + +- Section 3 – License Conditions. + + Your exercise of the Licensed Rights is expressly made subject to + the following conditions. + + - a. Attribution. + - 1. If You Share the Licensed Material (including in modified + form), You must: + - A. retain the following if it is supplied by the + Licensor with the Licensed Material: + - i. identification of the creator(s) of the Licensed + Material and any others designated to receive + attribution, in any reasonable manner requested by + the Licensor (including by pseudonym if designated); + - ii. a copyright notice; + - iii. a notice that refers to this Public License; + - iv. a notice that refers to the disclaimer of + warranties; + - v. a URI or hyperlink to the Licensed Material to + the extent reasonably practicable; + - B. indicate if You modified the Licensed Material and + retain an indication of any previous modifications; and + - C. indicate the Licensed Material is licensed under this + Public License, and include the text of, or the URI or + hyperlink to, this Public License. + - 2. You may satisfy the conditions in Section 3(a)(1) in any + reasonable manner based on the medium, means, and context in + which You Share the Licensed Material. For example, it may + be reasonable to satisfy the conditions by providing a URI + or hyperlink to a resource that includes the required + information. + - 3. If requested by the Licensor, You must remove any of the + information required by Section 3(a)(1)(A) to the extent + reasonably practicable. + - 4. If You Share Adapted Material You produce, the Adapter's + License You apply must not prevent recipients of the Adapted + Material from complying with this Public License. + +- Section 4 – Sui Generis Database Rights. + + Where the Licensed Rights include Sui Generis Database Rights that + apply to Your use of the Licensed Material: + + - a. for the avoidance of doubt, Section 2(a)(1) grants You the + right to extract, reuse, reproduce, and Share all or a + substantial portion of the contents of the database for + NonCommercial purposes only; + - b. if You include all or a substantial portion of the database + contents in a database in which You have Sui Generis Database + Rights, then the database in which You have Sui Generis Database + Rights (but not its individual contents) is Adapted Material; + and + - c. You must comply with the conditions in Section 3(a) if You + Share all or a substantial portion of the contents of the + database. + + For the avoidance of doubt, this Section 4 supplements and does not + replace Your obligations under this Public License where the + Licensed Rights include other Copyright and Similar Rights. + +- Section 5 – Disclaimer of Warranties and Limitation of Liability. + + - a. Unless otherwise separately undertaken by the Licensor, to + the extent possible, the Licensor offers the Licensed Material + as-is and as-available, and makes no representations or + warranties of any kind concerning the Licensed Material, whether + express, implied, statutory, or other. This includes, without + limitation, warranties of title, merchantability, fitness for a + particular purpose, non-infringement, absence of latent or other + defects, accuracy, or the presence or absence of errors, whether + or not known or discoverable. Where disclaimers of warranties + are not allowed in full or in part, this disclaimer may not + apply to You. + - b. To the extent possible, in no event will the Licensor be + liable to You on any legal theory (including, without + limitation, negligence) or otherwise for any direct, special, + indirect, incidental, consequential, punitive, exemplary, or + other losses, costs, expenses, or damages arising out of this + Public License or use of the Licensed Material, even if the + Licensor has been advised of the possibility of such losses, + costs, expenses, or damages. Where a limitation of liability is + not allowed in full or in part, this limitation may not apply to + You. + - c. The disclaimer of warranties and limitation of liability + provided above shall be interpreted in a manner that, to the + extent possible, most closely approximates an absolute + disclaimer and waiver of all liability. + +- Section 6 – Term and Termination. + + - a. This Public License applies for the term of the Copyright and + Similar Rights licensed here. However, if You fail to comply + with this Public License, then Your rights under this Public + License terminate automatically. + - b. Where Your right to use the Licensed Material has terminated + under Section 6(a), it reinstates: + + - 1. automatically as of the date the violation is cured, + provided it is cured within 30 days of Your discovery of the + violation; or + - 2. upon express reinstatement by the Licensor. + + For the avoidance of doubt, this Section 6(b) does not affect + any right the Licensor may have to seek remedies for Your + violations of this Public License. + + - c. For the avoidance of doubt, the Licensor may also offer the + Licensed Material under separate terms or conditions or stop + distributing the Licensed Material at any time; however, doing + so will not terminate this Public License. + - d. Sections 1, 5, 6, 7, and 8 survive termination of this Public + License. + +- Section 7 – Other Terms and Conditions. + + - a. The Licensor shall not be bound by any additional or + different terms or conditions communicated by You unless + expressly agreed. + - b. Any arrangements, understandings, or agreements regarding the + Licensed Material not stated herein are separate from and + independent of the terms and conditions of this Public License. + +- Section 8 – Interpretation. + + - a. For the avoidance of doubt, this Public License does not, and + shall not be interpreted to, reduce, limit, restrict, or impose + conditions on any use of the Licensed Material that could + lawfully be made without permission under this Public License. + - b. To the extent possible, if any provision of this Public + License is deemed unenforceable, it shall be automatically + reformed to the minimum extent necessary to make it enforceable. + If the provision cannot be reformed, it shall be severed from + this Public License without affecting the enforceability of the + remaining terms and conditions. + - c. No term or condition of this Public License will be waived + and no failure to comply consented to unless expressly agreed to + by the Licensor. + - d. Nothing in this Public License constitutes or may be + interpreted as a limitation upon, or waiver of, any privileges + and immunities that apply to the Licensor or You, including from + the legal processes of any jurisdiction or authority. + +Creative Commons is not a party to its public licenses. Notwithstanding, +Creative Commons may elect to apply one of its public licenses to +material it publishes and in those instances will be considered the +"Licensor." The text of the Creative Commons public licenses is +dedicated to the public domain under the CC0 Public Domain Dedication. +Except for the limited purpose of indicating that material is shared +under a Creative Commons public license or as otherwise permitted by the +Creative Commons policies published at creativecommons.org/policies, +Creative Commons does not authorize the use of the trademark "Creative +Commons" or any other trademark or logo of Creative Commons without its +prior written consent including, without limitation, in connection with +any unauthorized modifications to any of its public licenses or any +other arrangements, understandings, or agreements concerning use of +licensed material. For the avoidance of doubt, this paragraph does not +form part of the public licenses. + +Creative Commons may be contacted at creativecommons.org. \ No newline at end of file diff --git a/QWENVL-PRIVATE/README.md b/QWENVL-PRIVATE/README.md new file mode 100644 index 0000000000000000000000000000000000000000..07c22a6614e86a34b86c8af216d81d1408152b56 --- /dev/null +++ b/QWENVL-PRIVATE/README.md @@ -0,0 +1,1015 @@ +# Qwen2.5-VL + + +

+ +

+ +

+ 💜 Qwen Chat   |   🤗 Hugging Face   |   🤖 ModelScope   |   📑 Blog   |   📚 Cookbooks   |   📑 Paper   +
+🖥️ Demo   |   💬 WeChat (微信)   |   🫨 Discord   |   📑 API   |   🖥️ PAI-DSW +

+ + + + +## Introduction +In the past five months since Qwen2-VL's release, numerous developers have built new models on the Qwen2-VL vision-language models, providing us with valuable feedback. During this period, we focused on building more useful vision-language models. Today, we are excited to introduce the latest addition to the Qwen family: Qwen2.5-VL. + + +#### Key Enhancements: + +* **Powerful Document Parsing Capabilities**: Upgrade text recognition to omnidocument parsing, excelling in processing multi-scene, multilingual, and various built-in (handwriting, tables, charts, chemical formulas, and music sheets) documents. + +* **Precise Object Grounding Across Formats**: Unlock improved accuracy in detecting, pointing, and counting objects, accommodating absolute coordinate and JSON formats for advanced spatial reasoning. + +* **Ultra-long Video Understanding and Fine-grained Video Grounding**: Extend native dynamic resolution to the temporal dimension, enhancing the ability to understand videos lasting hours while extracting event segments in seconds. + +* **Enhanced Agent Functionality for Computer and Mobile Devices**: Leverage advanced grounding, reasoning, and decision-making abilities, boosting the model with superior agent functionality on smartphones and computers. + + +#### Model Architecture Updates: + +* **Dynamic Resolution and Frame Rate Training for Video Understanding**: + +We extend dynamic resolution to the temporal dimension by adopting dynamic FPS sampling, enabling the model to comprehend videos at various sampling rates. Accordingly, we update mRoPE in the time dimension with IDs and absolute time alignment, enabling the model to learn temporal sequence and speed, and ultimately acquire the ability to pinpoint specific moments. + +

+ +

+ + +* **Streamlined and Efficient Vision Encoder** + +We enhance both training and inference speeds by strategically implementing window attention into the ViT. The ViT architecture is further optimized with SwiGLU and RMSNorm, aligning it with the structure of the Qwen2.5 LLM. + + +## News +* 2025.04.08: We provide the [code](https://github.com/QwenLM/Qwen2.5-VL/tree/main/qwen-vl-finetune) for fine-tuning Qwen2-VL and Qwen2.5-VL. +* 2025.03.25: We have released the [Qwen2.5-VL-32B](https://huggingface.co/Qwen/Qwen2.5-VL-32B-Instruct). It is smarter and its responses align more closely with human preferences. For more details, please check our [blog](https://qwenlm.github.io/blog/qwen2.5-vl-32b/)! +* 2025.02.20: we have released the [Qwen2.5-VL Technical Report](https://arxiv.org/abs/2502.13923). Alongside the report, we have also released AWQ-quantized models for Qwen2.5-VL in three different sizes: [3B](https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct-AWQ), [7B](https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct-AWQ) , and [72B](https://huggingface.co/Qwen/Qwen2.5-VL-72B-Instruct-AWQ) parameters. +* 2025.01.28: We have released the [Qwen2.5-VL series](https://huggingface.co/Qwen). For more details, please check our [blog](https://qwenlm.github.io/blog/qwen2.5-vl/)! +* 2024.12.25: We have released the [QvQ-72B-Preview](https://huggingface.co/Qwen/QVQ-72B-Preview). QvQ-72B-Preview is an experimental research model, focusing on enhancing visual reasoning capabilities. For more details, please check our [blog](https://qwenlm.github.io/blog/qvq-72b-preview/)! +* 2024.09.19: The instruction-tuned [Qwen2-VL-72B model](https://huggingface.co/Qwen/Qwen2-VL-72B-Instruct) and its quantized version [[AWQ](https://huggingface.co/Qwen/Qwen2-VL-72B-Instruct-AWQ), [GPTQ-Int4](https://huggingface.co/Qwen/Qwen2-VL-72B-Instruct-GPTQ-Int4), [GPTQ-Int8](https://huggingface.co/Qwen/Qwen2-VL-72B-Instruct-GPTQ-Int8)] are now available. We have also released the [Qwen2-VL paper](https://arxiv.org/pdf/2409.12191) simultaneously. +* 2024.08.30: We have released the [Qwen2-VL series](https://huggingface.co/collections/Qwen/qwen2-vl-66cee7455501d7126940800d). The 2B and 7B models are now available, and the 72B model for open source is coming soon. For more details, please check our [blog](https://qwenlm.github.io/blog/qwen2-vl/)! + + +## Performance + + + +| Dataset | Qwen2.5-VL-3B
([🤗](https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct)[🤖](https://modelscope.cn/models/qwen/Qwen2.5-VL-3B-Instruct)) | Qwen2.5-VL-7B
([🤗](https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct)[🤖](https://modelscope.cn/models/qwen/Qwen2.5-VL-7B-Instruct)) | Qwen2.5-VL-32B
([🤗](https://huggingface.co/Qwen/Qwen2.5-VL-32B-Instruct)[🤖](https://modelscope.cn/models/qwen/Qwen2.5-VL-32B-Instruct)) | Qwen2.5-VL-72B
([🤗](https://huggingface.co/Qwen/Qwen2.5-VL-72B-Instruct)[🤖](https://modelscope.cn/models/qwen/Qwen2.5-VL-72B-Instruct)) | Gemini-2 Flash | GPT-4o | Claude3.5 Sonnet | Qwen2-VL 72B | +|--------------------|--------|--------|--------|----------------|----------------|--------|------------------|--------------| +| MMMU | 53.1 | 58.6 | 70.0 | **70.2** | **70.7** | 70.3 | 70.4 | 64.5 | +| MMMU Pro | 31.6 | 38.3 |49.5|**51.1** | **57** | 54.5 | 54.7 | 46.2 | +| DocVQA | 93.9 | 95.7 | 94.8 |**96.4** | 92.1 | 91.1 | 95.2 | **96.5** | +| InfoVQA | 77.1 | 82.6 | 83.4| **87.3** | 77.8 | 80.7 | 74.3 | 84.5 | +| CC-OCR | 74.5 | 77.8 | 77.1 |**79.8** | 73.0 | 66.6 | 62.7 | 68.7 | +| OCRBenchV2 | 54.3/52.1 | 56.3/57.2 | 57.2/59.1| **61.5/63.7** | - | 46.5/32.3 | 45.2/39.6 | 47.8/46.1 | +| MegaBench | 28.9 |36.8 | - |**51.3** | **55.2** | 54.2 | 52.1 | 46.8 | +| MMStar | 55.8|63.9| 69.5 |**70.8** | 69.4 | 64.7 | 65.1 | 68.3 | +| MMBench1.1 | 81.5 | 84.3 | 84.6|**88.0** | 83.0 | 82.1 | 83.4 | 86.6 | +| MathVista | 62.3 | 68.2 | 74.7 |**74.8** | 73.1 | 63.8 | 65.4 | 70.5 | +| MathVision | 21.2 | 25.1 | **38.4** |38.1 | **41.3** | 30.4 | 38.3 | 25.9 | +| VideoMME | 61.5/67.6 | 65.1/71.6 | 70.5/77.9 |**73.3/79.1** | -/- | 71.9/77.2 | 60/62.9 | 71.2/77.8 | +| MMBench-Video | 1.63 | 1.79 | 1.93 |**2.02** | - | 1.68 | 1.38 | 1.7 | +| LVBench | 43.3 | 45.3 | **49.0** | 47.3 | - | 30.8 | - | - | +| CharadesSTA | 38.8 | 43.6 | **54.2** | 50.9 | - | 35.7 | - | - | +| AITZ | 76.9|81.9 | 83.1 |**83.2** | - | 35.3 | - | - | +| Android Control | 63.7/90.8 | 60.1/91.4 | 69.6/93.3 |**67.36/93.7** | - | - | - | 66.4/84.4 | +| ScreenSpot | 55.5|84.7| **88.5** | 87.1 | 84.0 | 18.1 | 83.0 | - | +| ScreenSpot Pro | 23.9|29.0 | 39.4 |**43.6** | - | - | 17.1 | - | +| AndroidWorld | - | - | 22.0 |**35** | - | 34.5(SoM) | 27.9 | - | +| OSWorld | - | - | 5.92 |**8.83** | - | 5.03 | **14.9** | - | + + + + + +## Quickstart + +Below, we provide simple examples to show how to use Qwen2.5-VL with 🤖 ModelScope and 🤗 Transformers. + +The code of Qwen2.5-VL has been in the latest Hugging face transformers and we advise you to build from source with command: +``` +pip install transformers==4.51.3 accelerate +``` +or you might encounter the following error: +``` +KeyError: 'qwen2_5_vl' +``` + + +We offer a toolkit to help you handle various types of visual input more conveniently, as if you were using an API. This includes base64, URLs, and interleaved images and videos. You can install it using the following command: + +```bash +# It's highly recommended to use `[decord]` feature for faster video loading. +pip install qwen-vl-utils[decord] +``` + +Currently, `qwen-vl-utils` supports three video decoding backends: `torchvision`, `decord`, and `torchcodec`. While `decord` and `torchcodec` generally offer significantly faster decoding speeds compared to `torchvision`, we recommend using `torchcodec`. This is because `decord` has known issues, such as decoding hangs, and its project is no longer actively maintained. + +- For `decord`, if you are not using Linux, you might not be able to install `decord` from PyPI. In that case, you can use `pip install qwen-vl-utils` which will fall back to using torchvision for video processing. However, you can still [install decord from source](https://github.com/dmlc/decord?tab=readme-ov-file#install-from-source) to get decord used when loading video. + +- To use `torchcodec` as the backend for video decoding, follow the installation instructions provided in the official [torchcodec repository](https://github.com/pytorch/torchcodec/tree/main?tab=readme-ov-file#installing-torchcodec) and install it manually. Note that `torchcodec` depends on FFmpeg for decoding functionality. + +## Cookbooks + +We are preparing [cookbooks](https://github.com/QwenLM/Qwen2.5-VL/tree/main/cookbooks) for many capabilities, including recognition, localization, document parsing, video understanding, key information extraction, and more. Welcome to learn more! + +| Cookbook | Description | Open | +| -------- | ----------- | ---- | +| [Universal Recognition](https://github.com/QwenLM/Qwen2.5-VL/blob/main/cookbooks/universal_recognition.ipynb) | Not only identify animals, plants, people, and scenic spots but also recognize various objects such as cars and merchandise. | [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/QwenLM/Qwen2.5-VL/blob/main/cookbooks/universal_recognition.ipynb) | +| [Powerful Document Parsing Capabilities](https://github.com/QwenLM/Qwen2.5-VL/blob/main/cookbooks/document_parsing.ipynb) | The parsing of documents has reached a higher level, including not only text but also layout position information and our Qwen HTML format. | [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/QwenLM/Qwen2.5-VL/blob/main/cookbooks/document_parsing.ipynb) | +| [Precise Object Grounding Across Formats](https://github.com/QwenLM/Qwen2.5-VL/blob/main/cookbooks/spatial_understanding.ipynb) | Using absolute position coordinates, it supports both boxes and points, allowing for diverse combinations of positioning and labeling tasks. | [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/QwenLM/Qwen2.5-VL/blob/main/cookbooks/spatial_understanding.ipynb) | +| [General OCR and Key Information Extraction](https://github.com/QwenLM/Qwen2.5-VL/blob/main/cookbooks/ocr.ipynb) | Stronger text recognition capabilities in natural scenes and multiple languages, supporting diverse key information extraction needs. | [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/QwenLM/Qwen2.5-VL/blob/main/cookbooks/ocr.ipynb) | +| [Video Understanding](https://github.com/QwenLM/Qwen2.5-VL/blob/main/cookbooks/video_understanding.ipynb) | Better video OCR, long video understanding, and video grounding. | [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/QwenLM/Qwen2.5-VL/blob/main/cookbooks/video_understanding.ipynb) | +| [Mobile Agent](https://github.com/QwenLM/Qwen2.5-VL/blob/main/cookbooks/mobile_agent.ipynb) | Locate and think for mobile phone control. | [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/QwenLM/Qwen2.5-VL/blob/main/cookbooks/mobile_agent.ipynb) | +| [Computer-Use Agent](https://github.com/QwenLM/Qwen2.5-VL/blob/main/cookbooks/computer_use.ipynb) | Locate and think for controlling computers and Web. | [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/QwenLM/Qwen2.5-VL/blob/main/cookbooks/computer_use.ipynb) | + + +### Using 🤗 Transformers to Chat + +Here we show a code snippet to show you how to use the chat model with `transformers` and `qwen_vl_utils`: + +```python +from transformers import Qwen2_5_VLForConditionalGeneration, AutoProcessor +from qwen_vl_utils import process_vision_info + +# default: Load the model on the available device(s) +model = Qwen2_5_VLForConditionalGeneration.from_pretrained( + "Qwen/Qwen2.5-VL-7B-Instruct", torch_dtype="auto", device_map="auto" +) + +# We recommend enabling flash_attention_2 for better acceleration and memory saving, especially in multi-image and video scenarios. +# model = Qwen2_5_VLForConditionalGeneration.from_pretrained( +# "Qwen/Qwen2.5-VL-7B-Instruct", +# torch_dtype=torch.bfloat16, +# attn_implementation="flash_attention_2", +# device_map="auto", +# ) + +# default processor +processor = AutoProcessor.from_pretrained("Qwen/Qwen2.5-VL-7B-Instruct") + +# The default range for the number of visual tokens per image in the model is 4-16384. +# You can set min_pixels and max_pixels according to your needs, such as a token range of 256-1280, to balance performance and cost. +# min_pixels = 256*28*28 +# max_pixels = 1280*28*28 +# processor = AutoProcessor.from_pretrained("Qwen/Qwen2.5-VL-7B-Instruct", min_pixels=min_pixels, max_pixels=max_pixels) + +messages = [ + { + "role": "user", + "content": [ + { + "type": "image", + "image": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg", + }, + {"type": "text", "text": "Describe this image."}, + ], + } +] + +# Preparation for inference +text = processor.apply_chat_template( + messages, tokenize=False, add_generation_prompt=True +) +image_inputs, video_inputs = process_vision_info(messages) +inputs = processor( + text=[text], + images=image_inputs, + videos=video_inputs, + padding=True, + return_tensors="pt", +) +inputs = inputs.to(model.device) + +# Inference: Generation of the output +generated_ids = model.generate(**inputs, max_new_tokens=128) +generated_ids_trimmed = [ + out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) +] +output_text = processor.batch_decode( + generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False +) +print(output_text) +``` + +

+Minimum VRAM requirements + +| Precision | Qwen2.5-VL-3B | Qwen2.5-VL-7B | Qwen2.5-VL-72B | +|-----------|------------| --------- | -------- | +| FP32 | 11.5 GB | 26.34 GB | 266.21 GB | +| BF16 | 5.75 GB | 13.17 GB | 133.11 GB | +| INT8 | 2.87 GB | 6.59 GB | 66.5 GB | +| INT4 | 1.44 GB | 3.29 GB | 33.28 GB | + +Note: The table above presents the theoretical minimum video memory requirements for inference with `transformers`; however, in practice, the actual memory usage is typically at least 1.2 times higher. For more information, see the linked resource [here](https://huggingface.co/docs/accelerate/main/en/usage_guides/model_size_estimator). +
+ + +
+Multi image inference + +```python +# Messages containing multiple images and a text query +messages = [ + { + "role": "user", + "content": [ + {"type": "image", "image": "file:///path/to/image1.jpg"}, + {"type": "image", "image": "file:///path/to/image2.jpg"}, + {"type": "text", "text": "Identify the similarities between these images."}, + ], + } +] + +# Preparation for inference +text = processor.apply_chat_template( + messages, tokenize=False, add_generation_prompt=True +) +image_inputs, video_inputs = process_vision_info(messages) +inputs = processor( + text=[text], + images=image_inputs, + videos=video_inputs, + padding=True, + return_tensors="pt", +) +inputs = inputs.to("cuda") + +# Inference +generated_ids = model.generate(**inputs, max_new_tokens=128) +generated_ids_trimmed = [ + out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) +] +output_text = processor.batch_decode( + generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False +) +print(output_text) +``` +
+ +
+Video inference + +```python +# Messages containing a images list as a video and a text query +messages = [ + { + "role": "user", + "content": [ + { + "type": "video", + "video": [ + "file:///path/to/frame1.jpg", + "file:///path/to/frame2.jpg", + "file:///path/to/frame3.jpg", + "file:///path/to/frame4.jpg", + ], + }, + {"type": "text", "text": "Describe this video."}, + ], + } +] + +# Messages containing a local video path and a text query +messages = [ + { + "role": "user", + "content": [ + { + "type": "video", + "video": "file:///path/to/video1.mp4", + "max_pixels": 360 * 420, + "fps": 1.0, + }, + {"type": "text", "text": "Describe this video."}, + ], + } +] + +# Messages containing a video url and a text query +messages = [ + { + "role": "user", + "content": [ + { + "type": "video", + "video": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen2-VL/space_woaudio.mp4", + "min_pixels": 4 * 28 * 28, + "max_pixels": 256 * 28 * 28, + "total_pixels": 20480 * 28 * 28, + }, + {"type": "text", "text": "Describe this video."}, + ], + } +] + +# Preparation for inference +text = processor.apply_chat_template( + messages, tokenize=False, add_generation_prompt=True +) +image_inputs, video_inputs, video_kwargs = process_vision_info(messages, return_video_kwargs=True) +inputs = processor( + text=[text], + images=image_inputs, + videos=video_inputs, + fps=fps, + padding=True, + return_tensors="pt", + **video_kwargs, +) +inputs = inputs.to("cuda") + +# Inference +generated_ids = model.generate(**inputs, max_new_tokens=128) +generated_ids_trimmed = [ + out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) +] +output_text = processor.batch_decode( + generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False +) +print(output_text) +``` + +#### Video URL compatibility + +Video URL compatibility is primarily determined by the version of the third-party library being used. For more details, refer to the table below. If you prefer not to use the default backend, you can switch it by setting `FORCE_QWENVL_VIDEO_READER` to `torchvision`, `decord`, or `torchcodec`. + +| Backend | HTTP | HTTPS | +|-------------|------|-------| +| torchvision >= 0.19.0 | ✅ | ✅ | +| torchvision < 0.19.0 | ❌ | ❌ | +| decord | ✅ | ❌ | +| torchcodec | ✅ | ✅ | + +#### Configuration for adjusting video resolution + +We recommend setting appropriate values for the `min_pixels` and `max_pixels` parameters based on available GPU memory and the specific application scenario to restrict the resolution of individual frames in the video. Alternatively, you can use the `total_pixels` parameter to limit the total number of tokens in the video (it is recommended to set this value below 24576 * 28 * 28 to avoid excessively long input sequences). For more details on parameter usage and processing logic, please refer to the `fetch_video` function in `qwen_vl_utils/vision_process.py`. + +
+ +
+Batch inference + +```python +# Sample messages for batch inference +messages1 = [ + { + "role": "user", + "content": [ + {"type": "image", "image": "file:///path/to/image1.jpg"}, + {"type": "image", "image": "file:///path/to/image2.jpg"}, + {"type": "text", "text": "What are the common elements in these pictures?"}, + ], + } +] +messages2 = [ + {"role": "system", "content": "You are a helpful assistant."}, + {"role": "user", "content": "Who are you?"}, +] +# Combine messages for batch processing +messages = [messages1, messages2] + +# Preparation for batch inference +texts = [ + processor.apply_chat_template(msg, tokenize=False, add_generation_prompt=True) + for msg in messages +] +image_inputs, video_inputs = process_vision_info(messages) +inputs = processor( + text=texts, + images=image_inputs, + videos=video_inputs, + padding=True, + return_tensors="pt", +) +inputs = inputs.to("cuda") + +# Batch Inference +generated_ids = model.generate(**inputs, max_new_tokens=128) +generated_ids_trimmed = [ + out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) +] +output_texts = processor.batch_decode( + generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False +) +print(output_texts) +``` +
+ +### 🤖 ModelScope +We strongly advise users especially those in mainland China to use ModelScope. `snapshot_download` can help you solve issues concerning downloading checkpoints. + +### More Usage Tips + +For input images, we support local files, base64, and URLs. For videos, we currently only support local files. + +```python +# You can directly insert a local file path, a URL, or a base64-encoded image into the position where you want in the text. +## Local file path +messages = [ + { + "role": "user", + "content": [ + {"type": "image", "image": "file:///path/to/your/image.jpg"}, + {"type": "text", "text": "Describe this image."}, + ], + } +] +## Image URL +messages = [ + { + "role": "user", + "content": [ + {"type": "image", "image": "http://path/to/your/image.jpg"}, + {"type": "text", "text": "Describe this image."}, + ], + } +] +## Base64 encoded image +messages = [ + { + "role": "user", + "content": [ + {"type": "image", "image": "data:image;base64,/9j/..."}, + {"type": "text", "text": "Describe this image."}, + ], + } +] +``` +#### Image Resolution for performance boost + +The model supports a wide range of resolution inputs. By default, it uses the native resolution for input, but higher resolutions can enhance performance at the cost of more computation. Users can set the minimum and maximum number of pixels to achieve an optimal configuration for their needs, such as a token count range of 256-1280, to balance speed and memory usage. + +```python +min_pixels = 256 * 28 * 28 +max_pixels = 1280 * 28 * 28 +processor = AutoProcessor.from_pretrained( + "Qwen/Qwen2.5-VL-7B-Instruct", min_pixels=min_pixels, max_pixels=max_pixels +) +``` + +Besides, We provide two methods for fine-grained control over the image size input to the model: + +1. Specify exact dimensions: Directly set `resized_height` and `resized_width`. These values will be rounded to the nearest multiple of 28. + +2. Define min_pixels and max_pixels: Images will be resized to maintain their aspect ratio within the range of min_pixels and max_pixels. + +```python +# resized_height and resized_width +messages = [ + { + "role": "user", + "content": [ + { + "type": "image", + "image": "file:///path/to/your/image.jpg", + "resized_height": 280, + "resized_width": 420, + }, + {"type": "text", "text": "Describe this image."}, + ], + } +] +# min_pixels and max_pixels +messages = [ + { + "role": "user", + "content": [ + { + "type": "image", + "image": "file:///path/to/your/image.jpg", + "min_pixels": 50176, + "max_pixels": 50176, + }, + {"type": "text", "text": "Describe this image."}, + ], + } +] +``` + +#### Add ids for Multiple Image Inputs +By default, images and video content are directly included in the conversation. When handling multiple images, it's helpful to add labels to the images and videos for better reference. Users can control this behavior with the following settings: +
+Add vision ids + +```python +conversation = [ + { + "role": "user", + "content": [{"type": "image"}, {"type": "text", "text": "Hello, how are you?"}], + }, + { + "role": "assistant", + "content": "I'm doing well, thank you for asking. How can I assist you today?", + }, + { + "role": "user", + "content": [ + {"type": "text", "text": "Can you describe these images and video?"}, + {"type": "image"}, + {"type": "image"}, + {"type": "video"}, + {"type": "text", "text": "These are from my vacation."}, + ], + }, + { + "role": "assistant", + "content": "I'd be happy to describe the images and video for you. Could you please provide more context about your vacation?", + }, + { + "role": "user", + "content": "It was a trip to the mountains. Can you see the details in the images and video?", + }, +] + +# default: +prompt_without_id = processor.apply_chat_template( + conversation, add_generation_prompt=True +) +# Excepted output: '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n<|im_start|>user\n<|vision_start|><|image_pad|><|vision_end|>Hello, how are you?<|im_end|>\n<|im_start|>assistant\nI'm doing well, thank you for asking. How can I assist you today?<|im_end|>\n<|im_start|>user\nCan you describe these images and video?<|vision_start|><|image_pad|><|vision_end|><|vision_start|><|image_pad|><|vision_end|><|vision_start|><|video_pad|><|vision_end|>These are from my vacation.<|im_end|>\n<|im_start|>assistant\nI'd be happy to describe the images and video for you. Could you please provide more context about your vacation?<|im_end|>\n<|im_start|>user\nIt was a trip to the mountains. Can you see the details in the images and video?<|im_end|>\n<|im_start|>assistant\n' + + +# add ids +prompt_with_id = processor.apply_chat_template( + conversation, add_generation_prompt=True, add_vision_id=True +) +# Excepted output: '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n<|im_start|>user\nPicture 1: <|vision_start|><|image_pad|><|vision_end|>Hello, how are you?<|im_end|>\n<|im_start|>assistant\nI'm doing well, thank you for asking. How can I assist you today?<|im_end|>\n<|im_start|>user\nCan you describe these images and video?Picture 2: <|vision_start|><|image_pad|><|vision_end|>Picture 3: <|vision_start|><|image_pad|><|vision_end|>Video 1: <|vision_start|><|video_pad|><|vision_end|>These are from my vacation.<|im_end|>\n<|im_start|>assistant\nI'd be happy to describe the images and video for you. Could you please provide more context about your vacation?<|im_end|>\n<|im_start|>user\nIt was a trip to the mountains. Can you see the details in the images and video?<|im_end|>\n<|im_start|>assistant\n' +``` +
+ +#### Flash-Attention 2 to speed up generation + +First, make sure to install the latest version of Flash Attention 2: + +```bash +pip install -U flash-attn --no-build-isolation +``` + +Also, you should have a hardware that is compatible with Flash-Attention 2. Read more about it in the official documentation of the [flash attention repository](https://github.com/Dao-AILab/flash-attention). FlashAttention-2 can only be used when a model is loaded in `torch.float16` or `torch.bfloat16`. + +To load and run a model using Flash Attention-2, simply add `attn_implementation="flash_attention_2"` when loading the model as follows: + +```python +from transformers import Qwen2_5_VLForConditionalGeneration + +model = Qwen2_5_VLForConditionalGeneration.from_pretrained( + "Qwen/Qwen2.5-VL-7B-Instruct", + torch_dtype=torch.bfloat16, + attn_implementation="flash_attention_2", +) +``` + +### Processing Long Texts + +The current `config.json` is set for context length up to 32,768 tokens. +To handle extensive inputs exceeding 32,768 tokens, we utilize [YaRN](https://arxiv.org/abs/2309.00071), a technique for enhancing model length extrapolation, ensuring optimal performance on lengthy texts. + +For supported frameworks, you could add the following to `config.json` to enable YaRN: + +``` +{ + ..., + "type": "yarn", + "mrope_section": [ + 16, + 24, + 24 + ], + "factor": 4, + "original_max_position_embeddings": 32768 +} +``` + +However, it should be noted that this method has a significant impact on the performance of temporal and spatial localization tasks, and is therefore not recommended for use. + +At the same time, for long video inputs, since MRoPE itself is more economical with ids, the max_position_embeddings can be directly modified to a larger value, such as 64k. + + +### Try Qwen2.5-VL-72B with API! + +To explore Qwen2.5-VL-72B, a more fascinating multimodal model, we encourage you to test our cutting-edge API service. Let's start the exciting journey right now! + +#### Installation +```bash +pip install dashscope +``` + +#### Examples +```python +import dashscope + + +dashscope.api_key = "your_api_key" + +messages = [{ + 'role': 'user', + 'content': [ + { + 'image': "https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg" + }, + { + 'text': 'What are in the image?' + }, + ] +}] + +response = dashscope.MultiModalConversation.call(model='qwen2.5-vl-72b-instruct', messages=messages) +print(response) +``` + +For more usage, please refer to the tutorial at [aliyun](https://help.aliyun.com/zh/model-studio/developer-reference/qwen-vl-api). + + + + +## Demo +### Web UI Example + +In this section, we provide instructions for users to build a web-based user interface (UI) demo. This UI demo allows users to interact with a predefined model or application through a web browser. Follow the steps below to get started. + +#### Installation + +Before you begin, ensure that you have the required dependencies installed on your system. You can install them by running the following command: + +```bash +pip install -r requirements_web_demo.txt +``` + +#### Running the Demo with FlashAttention-2 + +Once the required packages are installed, you can launch the web demo using the following command. This command will start a web server and provide you with a link to access the UI in your web browser. + +**Recommended**: For enhanced performance and efficiency, especially in multi-image and video processing scenarios, we strongly recommend using [FlashAttention-2](https://github.com/Dao-AILab/flash-attention). FlashAttention-2 provides significant improvements in memory usage and speed, making it ideal for handling large-scale models and data processing. + +To enable FlashAttention-2, use the following command: + +```bash +python web_demo_mm.py --flash-attn2 +``` + +This will load the model with FlashAttention-2 enabled. + +**Default Usage**: If you prefer to run the demo without FlashAttention-2 or if you do not specify the `--flash-attn2` option, the demo will load the model using the standard attention implementation: + +```bash +python web_demo_mm.py +``` + +After running the command, you’ll see a link generated in the terminal similar to this: + +``` +Running on local: http://127.0.0.1:7860/ +``` + +Copy this link and paste it into your browser to access the web UI, where you can interact with the model by inputting text, uploading images, or using any other provided functionalities. + +##### Running the Streaming Video Chat Demo +An experimental streaming video chat demo is also available in the ``web_demo_streaming`` directory. + +To run the streaming video chat demo, use the following command: + +```bash +cd web_demo_streaming/ +python app.py --flash-attn2 +``` + +If you prefer to run the demo without FlashAttention-2, use the following command: +```bash +cd web_demo_streaming/ +python app.py +``` + +This demo supports webcam/screen capture as its video input source. To support screen capture video input, we use code snippet from the following hugginface space: [gstaff/gradio-screen-recorder](https://huggingface.co/spaces/gstaff/gradio-screen-recorder/tree/main). + + +## Deployment + +We recommend using vLLM for fast Qwen2.5-VL deployment and inference. You need to install `vllm>0.7.2` to enable Qwen2.5-VL support. You can also use our [official docker image](#-docker). + +You can also check [vLLM official documentation](https://docs.vllm.ai/en/latest/serving/multimodal_inputs.html) for more details about online serving and offline inference. + +### Installation +```bash +pip install git+https://github.com/huggingface/transformers@f3f6c86582611976e72be054675e2bf0abb5f775 +pip install accelerate +pip install qwen-vl-utils +pip install 'vllm>0.7.2' + +``` +### Start an OpenAI API Service + +Run the command below to start an OpenAI-compatible API service: + +```bash +vllm serve Qwen/Qwen2.5-VL-7B-Instruct --port 8000 --host 0.0.0.0 --dtype bfloat16 --limit-mm-per-prompt image=5,video=5 +``` + +Then you can use the chat API as below (via curl or Python API): + +```bash +curl http://localhost:8000/v1/chat/completions \ + -H "Content-Type: application/json" \ + -d '{ + "model": "Qwen/Qwen2.5-VL-7B-Instruct", + "messages": [ + {"role": "system", "content": "You are a helpful assistant."}, + {"role": "user", "content": [ + {"type": "image_url", "image_url": {"url": "https://modelscope.oss-cn-beijing.aliyuncs.com/resource/qwen.png"}}, + {"type": "text", "text": "What is the text in the illustrate?"} + ]} + ] + }' +``` + +```python +from openai import OpenAI + +# Set OpenAI's API key and API base to use vLLM's API server. +openai_api_key = "EMPTY" +openai_api_base = "http://localhost:8000/v1" + +client = OpenAI( + api_key=openai_api_key, + base_url=openai_api_base, +) + +chat_response = client.chat.completions.create( + model="Qwen/Qwen2.5-VL-7B-Instruct", + messages=[ + {"role": "system", "content": "You are a helpful assistant."}, + { + "role": "user", + "content": [ + { + "type": "image_url", + "image_url": { + "url": "https://modelscope.oss-cn-beijing.aliyuncs.com/resource/qwen.png" + }, + }, + {"type": "text", "text": "What is the text in the illustrate?"}, + ], + }, + ], +) +print("Chat response:", chat_response) +``` + +You can also upload base64-encoded local images (see [OpenAI API protocol document](https://platform.openai.com/docs/guides/vision/uploading-base-64-encoded-images) for more details): +```python +import base64 +from openai import OpenAI +# Set OpenAI's API key and API base to use vLLM's API server. +openai_api_key = "EMPTY" +openai_api_base = "http://localhost:8000/v1" +client = OpenAI( + api_key=openai_api_key, + base_url=openai_api_base, +) +image_path = "/path/to/local/image.png" +with open(image_path, "rb") as f: + encoded_image = base64.b64encode(f.read()) +encoded_image_text = encoded_image.decode("utf-8") +base64_qwen = f"data:image;base64,{encoded_image_text}" +chat_response = client.chat.completions.create( + model="Qwen/Qwen2.5-VL-7B-Instruct", + messages=[ + {"role": "system", "content": "You are a helpful assistant."}, + { + "role": "user", + "content": [ + { + "type": "image_url", + "image_url": { + "url": base64_qwen + }, + }, + {"type": "text", "text": "What is the text in the illustrate?"}, + ], + }, + ], +) +print("Chat response:", chat_response) +``` + +For videos, you can use the chat API as follows： +```python +import base64 +import numpy as np +from PIL import Image +from io import BytesIO +from openai import OpenAI +from qwen_vl_utils import process_vision_info + + +# Set OpenAI's API key and API base to use vLLM's API server. +openai_api_key = "EMPTY" +openai_api_base = "http://localhost:8000/v1" + +client = OpenAI( + api_key=openai_api_key, + base_url=openai_api_base, +) + + +video_messages = [ + {"role": "system", "content": "You are a helpful assistant."}, + {"role": "user", "content": [ + {"type": "text", "text": "请用表格总结一下视频中的商品特点"}, + { + "type": "video", + "video": "https://duguang-labelling.oss-cn-shanghai.aliyuncs.com/qiansun/video_ocr/videos/50221078283.mp4", + "total_pixels": 20480 * 28 * 28, "min_pixels": 16 * 28 * 2, + 'fps': 3.0 # The default value is 2.0, but for demonstration purposes, we set it to 3.0. + }] + }, +] + + +def prepare_message_for_vllm(content_messages): + """ + The frame extraction logic for videos in `vLLM` differs from that of `qwen_vl_utils`. + Here, we utilize `qwen_vl_utils` to extract video frames, with the `media_typ`e of the video explicitly set to `video/jpeg`. + By doing so, vLLM will no longer attempt to extract frames from the input base64-encoded images. + """ + vllm_messages, fps_list = [], [] + for message in content_messages: + message_content_list = message["content"] + if not isinstance(message_content_list, list): + vllm_messages.append(message) + continue + + new_content_list = [] + for part_message in message_content_list: + if 'video' in part_message: + video_message = [{'content': [part_message]}] + image_inputs, video_inputs, video_kwargs = process_vision_info(video_message, return_video_kwargs=True) + assert video_inputs is not None, "video_inputs should not be None" + video_input = (video_inputs.pop()).permute(0, 2, 3, 1).numpy().astype(np.uint8) + fps_list.extend(video_kwargs.get('fps', [])) + + # encode image with base64 + base64_frames = [] + for frame in video_input: + img = Image.fromarray(frame) + output_buffer = BytesIO() + img.save(output_buffer, format="jpeg") + byte_data = output_buffer.getvalue() + base64_str = base64.b64encode(byte_data).decode("utf-8") + base64_frames.append(base64_str) + + part_message = { + "type": "video_url", + "video_url": {"url": f"data:video/jpeg;base64,{','.join(base64_frames)}"} + } + new_content_list.append(part_message) + message["content"] = new_content_list + vllm_messages.append(message) + return vllm_messages, {'fps': fps_list} + + +video_messages, video_kwargs = prepare_message_for_vllm(video_messages) +chat_response = client.chat.completions.create( + model="Qwen/Qwen2.5-VL-7B-Instruct", + messages=video_messages, + extra_body={ + "mm_processor_kwargs": video_kwargs + } +) +print("Chat response:", chat_response) +``` + +### Inference Locally + +You can also use vLLM to inference Qwen2.5-VL locally: + +```python +from transformers import AutoProcessor +from vllm import LLM, SamplingParams +from qwen_vl_utils import process_vision_info + +MODEL_PATH = "Qwen/Qwen2.5-VL-7B-Instruct" + +llm = LLM( + model=MODEL_PATH, + limit_mm_per_prompt={"image": 10, "video": 10}, +) + +sampling_params = SamplingParams( + temperature=0.1, + top_p=0.001, + repetition_penalty=1.05, + max_tokens=256, + stop_token_ids=[], +) + +image_messages = [ + {"role": "system", "content": "You are a helpful assistant."}, + { + "role": "user", + "content": [ + { + "type": "image", + "image": "https://modelscope.oss-cn-beijing.aliyuncs.com/resource/qwen.png", + "min_pixels": 224 * 224, + "max_pixels": 1280 * 28 * 28, + }, + {"type": "text", "text": "What is the text in the illustrate?"}, + ], + }, +] + + +# For video input, you can pass following values instead: +# "type": "video", +# "video": "