Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,9 @@ 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
+assets/cat_and_chicken.mp4 filter=lfs diff=lfs merge=lfs -text
+assets/pipeline.png filter=lfs diff=lfs merge=lfs -text
+assets/sora.mp4 filter=lfs diff=lfs merge=lfs -text
+assets/sora.png filter=lfs diff=lfs merge=lfs -text
+videollama2/serve/examples/1034346401.mp4 filter=lfs diff=lfs merge=lfs -text
+videollama2/serve/examples/sample_demo_1.mp4 filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,58 @@
+# Python
+__pycache__
+*.pyc
+*.egg-info
+dist
+
+# Log
+*.log
+*.log.*
+*.json
+*.jsonl
+log_dir*/
+temp*/
+
+# Data
+!**/alpaca-data-conversation.json
+
+# Editor
+.idea
+*.swp
+
+# Other
+.DS_Store
+3rd_parties
+
+# jupyter
+.ipynb_checkpoints
+*.ipynb
+
+# DevContainer
+!.devcontainer/*
+
+# Demo
+serve_images/
+temp/
+
+# data folder
+data/
+dataset/
+datasets/
+
+# training folder
+wandb
+ckpts*
+output
+output/
+checkpoints
+checkpoints/
+work_dirs*/
+
+# evaluation folder
+/eval
+/eval*
+
+# pretrained weights
+pretrained/
+publish_models/
+public_models/

LICENSE

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

README.md

@@ -1,12 +1,345 @@
 ---
-title: Verblaze
-emoji: 🌖
-colorFrom: red
-colorTo: yellow
+title: verblaze
+app_file: videollama2/serve/gradio_web_server.py
 sdk: gradio
-sdk_version: 4.44.0
-app_file: app.py
-pinned: false
+sdk_version: 3.50.0
 ---
+<p align="center">
+    <img src="https://github.com/DAMO-NLP-SG/VideoLLaMA2/blob/e7bc34e0e9a96d77947a75b54399d9f96ccf209d/assets/logo.png" width="150" style="margin-bottom: 0.2;"/>
+<p>
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+<h3 align="center"><a href="https://arxiv.org/abs/2406.07476" style="color:#9C276A">
+VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMs</a></h3>
+<h5 align="center"> If our project helps you, please give us a star ⭐ on GitHub to support us. 🙏🙏 </h2>
+
+<h5 align="center">
+
+[![hf_space](https://img.shields.io/badge/🤗-Demo-9C276A.svg)](https://huggingface.co/spaces/lixin4ever/VideoLLaMA2)
+[![hf_checkpoint](https://img.shields.io/badge/🤗-Checkpoints-9C276A.svg)](https://huggingface.co/collections/DAMO-NLP-SG/videollama-2-6669b6b6f0493188305c87ed)
+[![hf_data](https://img.shields.io/badge/🤗-MSVC-9C276A.svg)](https://huggingface.co/datasets/DAMO-NLP-SG/Multi-Source-Video-Captioning)
+[![arXiv](https://img.shields.io/badge/Arxiv-2406.07476-AD1C18.svg?logo=arXiv)](https://arxiv.org/abs/2406.07476) <br>
+[![License](https://img.shields.io/badge/License-Apache%202.0-yellow)](https://github.com/DAMO-NLP-SG/VideoLLaMA2/blob/main/LICENSE) 
+[![Hits](https://hits.seeyoufarm.com/api/count/incr/badge.svg?url=https%3A%2F%2Fgithub.com%2FDAMO-NLP-SG%2FVideoLLaMA2&count_bg=%2379C83D&title_bg=%23555555&icon=&icon_color=%23E7E7E7&title=Visitor&edge_flat=false)](https://hits.seeyoufarm.com)
+[![GitHub issues](https://img.shields.io/github/issues/DAMO-NLP-SG/VideoLLaMA2?color=critical&label=Issues)](https://github.com/DAMO-NLP-SG/VideoLLaMA2/issues?q=is%3Aopen+is%3Aissue)
+[![GitHub closed issues](https://img.shields.io/github/issues-closed/DAMO-NLP-SG/VideoLLaMA2?color=success&label=Issues)](https://github.com/DAMO-NLP-SG/VideoLLaMA2/issues?q=is%3Aissue+is%3Aclosed)  <br>
+
+</h5>
+
+[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/videollama-2-advancing-spatial-temporal/zero-shot-video-question-answer-on-egoschema-1)](https://paperswithcode.com/sota/zero-shot-video-question-answer-on-egoschema-1?p=videollama-2-advancing-spatial-temporal) <br>
+[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/videollama-2-advancing-spatial-temporal/video-question-answering-on-perception-test)](https://paperswithcode.com/sota/video-question-answering-on-perception-test?p=videollama-2-advancing-spatial-temporal) <br>
+[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/videollama-2-advancing-spatial-temporal/video-question-answering-on-mvbench)](https://paperswithcode.com/sota/video-question-answering-on-mvbench?p=videollama-2-advancing-spatial-temporal) <br>
+[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/videollama-2-advancing-spatial-temporal/zero-shot-video-question-answer-on-video-mme-1)](https://paperswithcode.com/sota/zero-shot-video-question-answer-on-video-mme-1?p=videollama-2-advancing-spatial-temporal) <br>
+[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/videollama-2-advancing-spatial-temporal/zero-shot-video-question-answer-on-video-mme)](https://paperswithcode.com/sota/zero-shot-video-question-answer-on-video-mme?p=videollama-2-advancing-spatial-temporal) <br>
+
+<details open><summary>💡 Some other multimodal-LLM projects from our team may interest you ✨. </summary><p>
+<!--  may -->
+
+> [**Video-LLaMA: An Instruction-tuned Audio-Visual Language Model for Video Understanding**](https://github.com/DAMO-NLP-SG/Video-LLaMA) <br>
+> Hang Zhang, Xin Li, Lidong Bing <br>
+[![github](https://img.shields.io/badge/-Github-black?logo=github)](https://github.com/DAMO-NLP-SG/Video-LLaMA)  [![github](https://img.shields.io/github/stars/DAMO-NLP-SG/Video-LLaMA.svg?style=social)](https://github.com/DAMO-NLP-SG/Video-LLaMA) [![arXiv](https://img.shields.io/badge/Arxiv-2306.02858-b31b1b.svg?logo=arXiv)](https://arxiv.org/abs/2306.02858) <br>
+
+> [**VCD: Mitigating Object Hallucinations in Large Vision-Language Models through Visual Contrastive Decoding**](https://arxiv.org/abs/2311.16922) <br>
+> Sicong Leng, Hang Zhang, Guanzheng Chen, Xin Li, Shijian Lu, Chunyan Miao, Lidong Bing <br>
+[![github](https://img.shields.io/badge/-Github-black?logo=github)](https://github.com/DAMO-NLP-SG/VCD)  [![github](https://img.shields.io/github/stars/DAMO-NLP-SG/VCD.svg?style=social)](https://github.com/DAMO-NLP-SG/VCD)  [![arXiv](https://img.shields.io/badge/Arxiv-2311.16922-b31b1b.svg?logo=arXiv)](https://arxiv.org/abs/2311.16922) <br>
+
+</p></details>
+
+<div align="center"><video src="https://github.com/DAMO-NLP-SG/VideoLLaMA2/assets/18526640/e0e7951c-f392-42ed-afad-b2c7984d3e38" width="800"></div>
+
+
+## 📰 News
+* **[2024.08.14]**   Release checkpoints of [VideoLLaMA2-72B-Base](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-72B-Base) and [VideoLLaMA2-72B](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-72B)
+* **[2024.07.30]**  Release checkpoints of [VideoLLaMA2-8x7B-Base](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-8x7B-Base) and [VideoLLaMA2-8x7B](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-8x7B).
+* **[2024.06.25]**  🔥🔥 As of Jun 25, our [VideoLLaMA2-7B-16F](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-7B-16F) is the **Top-1** ~7B-sized VideoLLM on the [MLVU Leaderboard](https://github.com/JUNJIE99/MLVU?tab=readme-ov-file#trophy-mini-leaderboard).
+* **[2024.06.18]**  🔥🔥 As of Jun 18, our [VideoLLaMA2-7B-16F](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-7B-16F) is the **Top-1** ~7B-sized VideoLLM on the [VideoMME Leaderboard](https://video-mme.github.io/home_page.html#leaderboard).
+* **[2024.06.17]**  👋👋 Update technical report with the latest results and the missing references. If you have works closely related to VideoLLaMA 2 but not mentioned in the paper, feel free to let us know.  
+* **[2024.06.14]**  🔥🔥 [Online Demo](https://huggingface.co/spaces/lixin4ever/VideoLLaMA2) is available.
+* **[2024.06.03]**  Release training, evaluation, and serving codes of VideoLLaMA 2.
+
+
+<img src="https://github.com/DAMO-NLP-SG/VideoLLaMA2/assets/18526640/b9faf24f-bdd2-4728-9385-acea17ea086d" width="800" />
+
+## 🛠️ Requirements and Installation
+Basic Dependencies:
+* Python >= 3.8
+* Pytorch >= 2.2.0
+* CUDA Version >= 11.8
+* transformers == 4.40.0 (for reproducing paper results)
+* tokenizers == 0.19.1
+
+**[Online Mode]** Install required packages (better for development):
+```bash
+git clone https://github.com/DAMO-NLP-SG/VideoLLaMA2
+cd VideoLLaMA2
+pip install -r requirements.txt
+pip install flash-attn==2.5.8 --no-build-isolation
+```
+
+**[Offline Mode]** Install VideoLLaMA2 as a Python package (better for direct use):
+```bash
+git clone https://github.com/DAMO-NLP-SG/VideoLLaMA2
+cd VideoLLaMA2
+pip install --upgrade pip  # enable PEP 660 support
+pip install -e .
+pip install flash-attn==2.5.8 --no-build-isolation
+```
+
+## 🚀 Main Results
+
+### Multi-Choice Video QA & Video Captioning
+<p><img src="https://github.com/user-attachments/assets/fbe3e3c2-b0f1-4e29-8b92-bc3611192909" width="800" "/></p>
+
+###  Open-Ended Video QA
+<p><img src="https://github.com/user-attachments/assets/cee2efe1-309e-4301-a217-e2a848799953" width="800" "/></p>
+
+
+
+## :earth_americas: Model Zoo
+| Model Name     | Model Type | Visual Encoder | Language Decoder | # Training Frames |
+|:----------------|:------------:|:----------------|:------------------|:----------------:|
+| [VideoLLaMA2-7B-Base](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-7B-Base)  | Base  | [clip-vit-large-patch14-336](https://huggingface.co/openai/clip-vit-large-patch14-336) | [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2)  | 8 |
+| [VideoLLaMA2-7B](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-7B)  | Chat | [clip-vit-large-patch14-336](https://huggingface.co/openai/clip-vit-large-patch14-336) | [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2)  | 8 |
+| [VideoLLaMA2-7B-16F-Base](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-7B-16F-Base)  | Base  | [clip-vit-large-patch14-336](https://huggingface.co/openai/clip-vit-large-patch14-336) | [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2)  | 16 |
+| [VideoLLaMA2-7B-16F](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-7B-16F)  | Chat | [clip-vit-large-patch14-336](https://huggingface.co/openai/clip-vit-large-patch14-336) | [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2)  | 16 |
+| [VideoLLaMA2-8x7B-Base](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-8x7B-Base)  | Base | [clip-vit-large-patch14-336](https://huggingface.co/openai/clip-vit-large-patch14-336) | [Mixtral-8x7B-Instruct-v0.1](https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1)  | 8 |
+| [VideoLLaMA2-8x7B](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-8x7B)  | Chat | [clip-vit-large-patch14-336](https://huggingface.co/openai/clip-vit-large-patch14-336) | [Mixtral-8x7B-Instruct-v0.1](https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1)  | 8 |
+| [VideoLLaMA2-72B-Base](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-72B-Base)  | Base | [clip-vit-large-patch14-336](https://huggingface.co/openai/clip-vit-large-patch14-336) | [Qwen2-72B-Instruct](https://huggingface.co/Qwen/Qwen2-72B-Instruct)  | 8 |
+| [VideoLLaMA2-72B](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-72B)  | Chat | [clip-vit-large-patch14-336](https://huggingface.co/openai/clip-vit-large-patch14-336) | [Qwen2-72B-Instruct](https://huggingface.co/Qwen/Qwen2-72B-Instruct)  | 8 |
+
+
+## [🤗 Demo](https://huggingface.co/spaces/lixin4ever/VideoLLaMA2)
+
+It is highly recommended to try our [online demo](https://huggingface.co/spaces/lixin4ever/VideoLLaMA2) first.
+
+To run a video-based LLM (Large Language Model) web demonstration on your device, you will first need to ensure that you have the necessary model checkpoints prepared, followed by adhering to the steps outlined to successfully launch the demo.
+
+### Single-model Version
+
+* Launch a gradio app directly ([VideoLLaMA2-7B](https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2-7B) is adopted by default):
+```bash
+python videollama2/serve/gradio_web_server_adhoc.py
+```
+
+### Multi-model Version
+
+1. Launch a global controller
+```bash
+cd /path/to/VideoLLaMA2
+python -m videollama2.serve.controller --host 0.0.0.0 --port 10000
+```
+
+2. Launch a gradio webserver
+```bash
+python -m videollama2.serve.gradio_web_server --controller http://localhost:10000 --model-list-mode reload
+```
+
+3. Launch one or multiple model workers
+```bash
+#  export HF_ENDPOINT=https://hf-mirror.com  # If you are unable to access Hugging Face, try to uncomment this line.
+python -m videollama2.serve.model_worker --host 0.0.0.0 --controller http://localhost:10000 --port 40000 --worker http://localhost:40000 --model-path /PATH/TO/MODEL1
+python -m videollama2.serve.model_worker --host 0.0.0.0 --controller http://localhost:10000 --port 40001 --worker http://localhost:40001 --model-path /PATH/TO/MODEL2
+python -m videollama2.serve.model_worker --host 0.0.0.0 --controller http://localhost:10000 --port 40002 --worker http://localhost:40002 --model-path /PATH/TO/MODEL3
+...
+```
+
+
+## 🗝️ Training & Evaluation
+
+### Quick Start
+
+To facilitate further development on top of our codebase, we provide a quick-start guide on how to train a customized [VideoLLaMA2](https://github.com/DAMO-NLP-SG/VideoLLaMA2) with [VideoLLaVA](https://github.com/PKU-YuanGroup/Video-LLaVA) dataset and evaluate the trained model on the mainstream video-llm benchmarks.
+
+1. Training Data Structure:
+```bash
+VideoLLaMA2
+├── datasets
+│   ├── videollava_pt
+|   |   ├── llava_image/ # Available at: https://pan.baidu.com/s/17GYcE69FcJjjUM0e4Gad2w?pwd=9ga3 or https://drive.google.com/drive/folders/1QmFj2FcMAoWNCUyiUtdcW0-IOhLbOBcf?usp=drive_link
+|   |   ├── valley/      # Available at: https://pan.baidu.com/s/1jluOimE7mmihEBfnpwwCew?pwd=jyjz or https://drive.google.com/drive/folders/1QmFj2FcMAoWNCUyiUtdcW0-IOhLbOBcf?usp=drive_link
+|   |   └── valley_llavaimage.json # Available at: https://drive.google.com/file/d/1zGRyVSUMoczGq6cjQFmT0prH67bu2wXD/view, including 703K video-text and 558K image-text pairs
+│   ├── videollava_sft
+|   |   ├── llava_image_tune/  # Available at: https://pan.baidu.com/s/1l-jT6t_DlN5DTklwArsqGw?pwd=o6ko
+|   |   ├── videochatgpt_tune/ # Available at: https://pan.baidu.com/s/10hJ_U7wVmYTUo75YHc_n8g?pwd=g1hf
+|   |   └── videochatgpt_llavaimage_tune.json # Available at: https://drive.google.com/file/d/1zGRyVSUMoczGq6cjQFmT0prH67bu2wXD/view, including 100K video-centric, 625K image-centric and 40K text-only conversations
+```
+2. Command:
+```bash
+# VideoLLaMA2-vllava pretraining
+bash scripts/vllava/pretrain.sh
+# VideoLLaMA2-vllava finetuning
+bash scripts/vllava/finetune.sh
+```
+3. Evaluation Data Structure:
+```bash
+VideoLLaMA2
+├── eval
+│   ├── egoschema # Official website: https://github.com/egoschema/EgoSchema
+|   |   ├── good_clips_git/ # Available at: https://drive.google.com/drive/folders/1SS0VVz8rML1e5gWq7D7VtP1oxE2UtmhQ
+|   |   └── questions.json  # Available at: https://github.com/egoschema/EgoSchema/blob/main/questions.json
+│   ├── mvbench # Official website: https://huggingface.co/datasets/OpenGVLab/MVBench
+|   |   ├── video/
+|   |   |   ├── clever/
+|   |   |   └── ...
+|   |   └── json/
+|   |   |   ├── action_antonym.json
+|   |   |   └── ...
+│   ├── perception_test_mcqa # Official website: https://huggingface.co/datasets/OpenGVLab/MVBench
+|   |   ├── videos/ # Available at: https://storage.googleapis.com/dm-perception-test/zip_data/test_videos.zip
+|   |   └── mc_question_test.json # Download from https://storage.googleapis.com/dm-perception-test/zip_data/mc_question_test_annotations.zip
+│   ├── videomme # Official website: https://video-mme.github.io/home_page.html#leaderboard
+|   |   ├── test-00000-of-00001.parquet
+|   |   ├── videos/
+|   |   └── subtitles/
+│   ├── Activitynet_Zero_Shot_QA # Official website: https://github.com/MILVLG/activitynet-qa
+|   |   ├── all_test/   # Available at: https://mbzuaiac-my.sharepoint.com/:u:/g/personal/hanoona_bangalath_mbzuai_ac_ae/EatOpE7j68tLm2XAd0u6b8ABGGdVAwLMN6rqlDGM_DwhVA?e=90WIuW
+|   |   ├── test_q.json # Available at: https://github.com/MILVLG/activitynet-qa/tree/master/dataset
+|   |   └── test_a.json # Available at: https://github.com/MILVLG/activitynet-qa/tree/master/dataset
+│   ├── MSVD_Zero_Shot_QA # Official website: https://github.com/xudejing/video-question-answering
+|   |   ├── videos/     
+|   |   ├── test_q.json 
+|   |   └── test_a.json
+│   ├── videochatgpt_gen # Official website: https://github.com/mbzuai-oryx/Video-ChatGPT/tree/main/quantitative_evaluation
+|   |   ├── Test_Videos/ # Available at: https://mbzuaiac-my.sharepoint.com/:u:/g/personal/hanoona_bangalath_mbzuai_ac_ae/EatOpE7j68tLm2XAd0u6b8ABGGdVAwLMN6rqlDGM_DwhVA?e=90WIuW
+|   |   ├── Test_Human_Annotated_Captions/ # Available at: https://mbzuaiac-my.sharepoint.com/personal/hanoona_bangalath_mbzuai_ac_ae/_layouts/15/onedrive.aspx?id=%2Fpersonal%2Fhanoona%5Fbangalath%5Fmbzuai%5Fac%5Fae%2FDocuments%2FVideo%2DChatGPT%2FData%5FCode%5FModel%5FRelease%2FQuantitative%5FEvaluation%2Fbenchamarking%2FTest%5FHuman%5FAnnotated%5FCaptions%2Ezip&parent=%2Fpersonal%2Fhanoona%5Fbangalath%5Fmbzuai%5Fac%5Fae%2FDocuments%2FVideo%2DChatGPT%2FData%5FCode%5FModel%5FRelease%2FQuantitative%5FEvaluation%2Fbenchamarking&ga=1
+|   |   ├── generic_qa.json     # These three json files available at: https://mbzuaiac-my.sharepoint.com/personal/hanoona_bangalath_mbzuai_ac_ae/_layouts/15/onedrive.aspx?id=%2Fpersonal%2Fhanoona%5Fbangalath%5Fmbzuai%5Fac%5Fae%2FDocuments%2FVideo%2DChatGPT%2FData%5FCode%5FModel%5FRelease%2FQuantitative%5FEvaluation%2Fbenchamarking%2FBenchmarking%5FQA&ga=1
+|   |   ├── temporal_qa.json
+|   |   └── consistency_qa.json
+```
+4. Command:
+```bash
+# mvbench evaluation
+CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 bash scripts/eval/eval_video_qa_mvbench.sh
+# activitynet-qa evaluation (need to set azure openai key/endpoint/deployname)
+CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 bash scripts/eval/eval_video_qa_mvbench.sh
+```
+
+### Data Format
+
+If you want to train a video-llm on your data, you need to follow the procedures below to prepare the video/image sft data:
+
+1. Suppose your data structure is like:
+```bash
+VideoLLaMA2
+├── datasets
+│   ├── custom_sft
+│   |   ├── images
+│   |   ├── videos
+|   |   └── custom.json
+```
+2. Then you should re-organize the annotated video/image sft data according to the following format:
+```json
+[
+    {
+        "id": 0,
+        "video": "images/xxx.jpg",
+        "conversations": [
+            {
+                "from": "human",
+                "value": "<image>\nWhat are the colors of the bus in the image?"
+            },
+            {
+                "from": "gpt",
+                "value": "The bus in the image is white and red."
+            },
+            ...
+        ],
+    }
+    {
+        "id": 1,
+        "video": "videos/xxx.mp4",
+        "conversations": [
+            {
+                "from": "human",
+                "value": "<video>\nWhat are the main activities that take place in the video?"
+            },
+            {
+                "from": "gpt",
+                "value": "The main activities that take place in the video are the preparation of camera equipment by a man, a group of men riding a helicopter, and a man sailing a boat through the water."
+            },
+            ...
+        ],
+    },
+    ...
+]
+```
+3. Modify the `scripts/custom/finetune.sh`:
+```bash
+...
+--data_path datasets/custom_sft/custom.json
+--data_folder datasets/custom_sft/
+--pretrain_mm_mlp_adapter CONNECTOR_DOWNLOAD_PATH (e.g., DAMO-NLP-SG/VideoLLaMA2-7B-Base)
+...
+```
+
+## 🤖 Inference
+
+Video/Image Inference:
+```python
+import sys
+sys.path.append('./')
+from videollama2 import model_init, mm_infer
+from videollama2.utils import disable_torch_init
+
+
+def inference():
+    disable_torch_init()
+
+    # Video Inference
+    modal = 'video'
+    modal_path = 'assets/cat_and_chicken.mp4' 
+    instruct = 'What animals are in the video, what are they doing, and how does the video feel?'
+    # Reply:
+    # The video features a kitten and a baby chick playing together. The kitten is seen laying on the floor while the baby chick hops around. The two animals interact playfully with each other, and the video has a cute and heartwarming feel to it.
+
+    # Image Inference
+    modal = 'image'
+    modal_path = 'assets/sora.png'
+    instruct = 'What is the woman wearing, what is she doing, and how does the image feel?'
+    # Reply:
+    # The woman in the image is wearing a black coat and sunglasses, and she is walking down a rain-soaked city street. The image feels vibrant and lively, with the bright city lights reflecting off the wet pavement, creating a visually appealing atmosphere. The woman's presence adds a sense of style and confidence to the scene, as she navigates the bustling urban environment.
+
+    model_path = 'DAMO-NLP-SG/VideoLLaMA2-7B'
+    # Base model inference (only need to replace model_path)
+    # model_path = 'DAMO-NLP-SG/VideoLLaMA2-7B-Base'
+    model, processor, tokenizer = model_init(model_path)
+    output = mm_infer(processor[modal](modal_path), instruct, model=model, tokenizer=tokenizer, do_sample=False, modal=modal)
+
+    print(output)
+
+if __name__ == "__main__":
+    inference()
+```
+
+## 📑 Citation
+
+If you find VideoLLaMA useful for your research and applications, please cite using this BibTeX:
+```bibtex
+@article{damonlpsg2024videollama2,
+  title={VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMs},
+  author={Cheng, Zesen and Leng, Sicong and Zhang, Hang and Xin, Yifei and Li, Xin and Chen, Guanzheng and Zhu, Yongxin and Zhang, Wenqi and Luo, Ziyang and Zhao, Deli and Bing, Lidong},
+  journal={arXiv preprint arXiv:2406.07476},
+  year={2024},
+  url = {https://arxiv.org/abs/2406.07476}
+}
+
+@article{damonlpsg2023videollama,
+  title = {Video-LLaMA: An Instruction-tuned Audio-Visual Language Model for Video Understanding},
+  author = {Zhang, Hang and Li, Xin and Bing, Lidong},
+  journal = {arXiv preprint arXiv:2306.02858},
+  year = {2023},
+  url = {https://arxiv.org/abs/2306.02858}
+}
+```
+
+## 👍 Acknowledgement
+The codebase of VideoLLaMA 2 is adapted from [**LLaVA 1.5**](https:github.com/haotian-liu/LLaVA) and [**FastChat**](https://github.com/lm-sys/FastChat). We are also grateful for the following projects our VideoLLaMA 2 arise from:
+* [**LLaMA 2**](https://github.com/meta-llama/llama), [**Mistral-7B**](https://mistral.ai/news/announcing-mistral-7b/), [**OpenAI CLIP**](https://openai.com/index/clip/), [**Honeybee**](https://github.com/kakaobrain/honeybee).
+* [**Video-ChatGPT**](https://github.com/mbzuai-oryx/Video-ChatGPT), [**Video-LLaVA**](https://github.com/PKU-YuanGroup/Video-LLaVA). 
+* [**WebVid**](https://github.com/m-bain/webvid), [**Panda-70M**](https://github.com/snap-research/Panda-70M), [**LanguageBind**](https://github.com/PKU-YuanGroup/LanguageBind), [**InternVid**](https://github.com/OpenGVLab/InternVideo/tree/main/Data/InternVid).
+* [**VideoChat2**](https://github.com/OpenGVLab/Ask-Anything/tree/main/video_chat2), [**Valley**](https://github.com/RupertLuo/Valley), [**VTimeLLM**](https://github.com/huangb23/VTimeLLM), [**ShareGPT4V**](https://sharegpt4v.github.io/).
+
+
+## 🔒 License
+
+This project is released under the Apache 2.0 license as found in the LICENSE file.
+The service is a research preview intended for **non-commercial use ONLY**, subject to the model Licenses of LLaMA and Mistral, Terms of Use of the data generated by OpenAI, and Privacy Practices of ShareGPT. Please get in touch with us if you find any potential violations.

assets/cat_and_chicken.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f24723064ee27ea8fc7a30b4542601ed03a42952c0d20fe918213cf876bfec4
+size 18956323

assets/sora.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24e5f0ea3353f23225d00efcdf136fa6dc346301fc34082790e2152c80fa0490
+size 14978533

aws/README.md

@@ -0,0 +1,54 @@
+# AWS CLI v2
+
+This bundle contains a built executable of the AWS CLI v2.
+
+## Installation
+
+To install the AWS CLI v2, run the `install` script:
+```
+$ sudo ./install 
+You can now run: /usr/local/bin/aws --version
+```
+This will install the AWS CLI v2 at `/usr/local/bin/aws`.  Assuming
+`/usr/local/bin` is on your `PATH`, you can now run:
+```
+$ aws --version
+```
+
+
+### Installing without sudo
+
+If you don't have ``sudo`` permissions or want to install the AWS
+CLI v2 only for the current user, run the `install` script with the `-b`
+and `-i` options:
+```
+$ ./install -i ~/.local/aws-cli -b ~/.local/bin
+``` 
+This will install the AWS CLI v2 in `~/.local/aws-cli` and create
+symlinks for `aws` and `aws_completer` in `~/.local/bin`. For more
+information about these options, run the `install` script with `-h`:
+```
+$ ./install -h
+```
+
+### Updating
+
+If you run the `install` script and there is a previously installed version
+of the AWS CLI v2, the script will error out. To update to the version included
+in this bundle, run the `install` script with `--update`:
+```
+$ sudo ./install --update
+```
+
+
+### Removing the installation
+
+To remove the AWS CLI v2, delete the its installation and symlinks:
+```
+$ sudo rm -rf /usr/local/aws-cli
+$ sudo rm /usr/local/bin/aws
+$ sudo rm /usr/local/bin/aws_completer
+```
+Note if you installed the AWS CLI v2 using the `-b` or `-i` options, you will
+need to remove the installation and the symlinks in the directories you
+specified.

aws/THIRD_PARTY_LICENSES

@@ -0,0 +1,1468 @@
+** cryptography 3.3.2; version 3.3.2 --
+https://github.com/pyca/cryptography/tree/3.3.2
+Copyright (c) Individual contributors.
+All rights reserved.
+
+This software is made available under the terms of *either* of the licenses
+found in LICENSE.APACHE or LICENSE.BSD. Contributions to cryptography are made
+under the terms of *both* these licenses.
+
+The code used in the OpenSSL locking callback and OS random engine is derived
+from CPython, and is licensed under the terms of the PSF License Agreement.
+
+* For cryptography 3.3.2 see also this required NOTICE:
+    Copyright (c) Individual contributors.
+    All rights reserved.
+
+------
+
+** botocore; version 2 -- https://github.com/boto/botocore/tree/v2
+Botocore
+Copyright 2012-2017 Amazon.com, Inc. or its affiliates. All Rights Reserved.
+
+----
+
+Botocore includes a vendorized copy of the requests python library to ease
+installation.
+
+Requests License
+================
+
+Copyright 2013 Kenneth Reitz
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.
+
+
+The requests library also includes some vendorized python libraries to ease
+installation.
+
+Urllib3 License
+===============
+
+This is the MIT license: http://www.opensource.org/licenses/mit-license.php
+
+Copyright 2008-2011 Andrey Petrov and contributors (see CONTRIBUTORS.txt),
+Modifications copyright 2012 Kenneth Reitz.
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of
+this
+software and associated documentation files (the "Software"), to deal in the
+Software
+without restriction, including without limitation the rights to use, copy,
+modify, merge,
+publish, distribute, sublicense, and/or sell copies of the Software, and to
+permit persons
+to whom the Software is furnished to do so, subject to the following
+conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or
+substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED,
+INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A
+PARTICULAR
+PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+BE LIABLE
+FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+TORT OR
+OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE
+OR OTHER
+DEALINGS IN THE SOFTWARE.
+
+Chardet License
+===============
+
+This library is free software; you can redistribute it and/or
+modify it under the terms of the GNU Lesser General Public
+License as published by the Free Software Foundation; either
+version 2.1 of the License, or (at your option) any later version.
+
+This library is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+Lesser General Public License for more details.
+
+You should have received a copy of the GNU Lesser General Public
+License along with this library; if not, write to the Free Software
+Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
+02110-1301  USA
+
+Bundle of CA Root Certificates
+==============================
+
+This library is free software; you can redistribute it and/or
+modify it under the terms of the GNU Lesser General Public
+License as published by the Free Software Foundation; either
+version 2.1 of the License, or (at your option) any later version.
+
+This library is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+Lesser General Public License for more details.
+
+You should have received a copy of the GNU Lesser General Public
+License along with this library; if not, write to the Free Software
+Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
+02110-1301
+** s3transfer; version 0 -- https://github.com/boto/s3transfer
+s3transfer
+Copyright 2016 Amazon.com, Inc. or its affiliates. All Rights Reserved.
+
+Apache License
+
+Version 2.0, January 2004
+
+http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND
+DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction, and
+      distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by the
+      copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all other
+      entities that control, are controlled by, or are under common control
+      with that entity. For the purposes of this definition, "control" means
+      (i) the power, direct or indirect, to cause the direction or management
+      of such entity, whether by contract or otherwise, or (ii) ownership of
+      fifty percent (50%) or more of the outstanding shares, or (iii)
+      beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity exercising
+      permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation source,
+      and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but not limited
+      to compiled object code, generated documentation, and conversions to
+      other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or Object
+      form, made available under the License, as indicated by a copyright
+      notice that is included in or attached to the work (an example is
+      provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object form,
+      that is based on (or derived from) the Work and for which the editorial
+      revisions, annotations, elaborations, or other modifications represent,
+      as a whole, an original work of authorship. For the purposes of this
+      License, Derivative Works shall not include works that remain separable
+      from, or merely link (or bind by name) to the interfaces of, the Work and
+      Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including the original
+      version of the Work and any modifications or additions to that Work or
+      Derivative Works thereof, that is intentionally submitted to Licensor for
+      inclusion in the Work by the copyright owner or by an individual or Legal
+      Entity authorized to submit on behalf of the copyright owner. For the
+      purposes of this definition, "submitted" means any form of electronic,
+      verbal, or written communication sent to the Licensor or its
+      representatives, including but not limited to communication on electronic
+      mailing lists, source code control systems, and issue tracking systems
+      that are managed by, or on behalf of, the Licensor for the purpose of
+      discussing and improving the Work, but excluding communication that is
+      conspicuously marked or otherwise designated in writing by the copyright
+      owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity on
+      behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of this
+   License, each Contributor hereby grants to You a perpetual, worldwide,
+   non-exclusive, no-charge, royalty-free, irrevocable copyright license to
+   reproduce, prepare Derivative Works of, publicly display, publicly perform,
+   sublicense, and distribute the Work and such Derivative Works in Source or
+   Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of this
+   License, each Contributor hereby grants to You a perpetual, worldwide,
+   non-exclusive, no-charge, royalty-free, irrevocable (except as stated in
+   this section) patent license to make, have made, use, offer to sell, sell,
+   import, and otherwise transfer the Work, where such license applies only to
+   those patent claims licensable by such Contributor that are necessarily
+   infringed by their Contribution(s) alone or by combination of their
+   Contribution(s) with the Work to which such Contribution(s) was submitted.
+   If You institute patent litigation against any entity (including a
+   cross-claim or counterclaim in a lawsuit) alleging that the Work or a
+   Contribution incorporated within the Work constitutes direct or contributory
+   patent infringement, then any patent licenses granted to You under this
+   License for that Work shall terminate as of the date such litigation is
+   filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the Work or
+   Derivative Works thereof in any medium, with or without modifications, and
+   in Source or Object form, provided that You meet the following conditions:
+
+      (a) You must give any other recipients of the Work or Derivative Works a
+      copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices stating
+      that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works that You
+      distribute, all copyright, patent, trademark, and attribution notices
+      from the Source form of the Work, excluding those notices that do not
+      pertain to any part of the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+      distribution, then any Derivative Works that You distribute must include
+      a readable copy of the attribution notices contained within such NOTICE
+      file, excluding those notices that do not pertain to any part of the
+      Derivative Works, in at least one of the following places: within a
+      NOTICE text file distributed as part of the Derivative Works; within the
+      Source form or documentation, if provided along with the Derivative
+      Works; or, within a display generated by the Derivative Works, if and
+      wherever such third-party notices normally appear. The contents of the
+      NOTICE file are for informational purposes only and do not modify the
+      License. You may add Your own attribution notices within Derivative Works
+      that You distribute, alongside or as an addendum to the NOTICE text from
+      the Work, provided that such additional attribution notices cannot be
+      construed as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and may
+      provide additional or different license terms and conditions for use,
+      reproduction, or distribution of Your modifications, or for any such
+      Derivative Works as a whole, provided Your use, reproduction, and
+      distribution of the Work otherwise complies with the conditions stated in
+      this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise, any
+   Contribution intentionally submitted for inclusion in the Work by You to the
+   Licensor shall be under the terms and conditions of this License, without
+   any additional terms or conditions. Notwithstanding the above, nothing
+   herein shall supersede or modify the terms of any separate license agreement
+   you may have executed with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+   names, trademarks, service marks, or product names of the Licensor, except
+   as required for reasonable and customary use in describing the origin of the
+   Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or agreed to in
+   writing, Licensor provides the Work (and each Contributor provides its
+   Contributions) on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+   KIND, either express or implied, including, without limitation, any
+   warranties or conditions of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or
+   FITNESS FOR A PARTICULAR PURPOSE. You are solely responsible for determining
+   the appropriateness of using or redistributing the Work and assume any risks
+   associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory, whether
+   in tort (including negligence), contract, or otherwise, unless required by
+   applicable law (such as deliberate and grossly negligent acts) or agreed to
+   in writing, shall any Contributor be liable to You for damages, including
+   any direct, indirect, special, incidental, or consequential damages of any
+   character arising as a result of this License or out of the use or inability
+   to use the Work (including but not limited to damages for loss of goodwill,
+   work stoppage, computer failure or malfunction, or any and all other
+   commercial damages or losses), even if such Contributor has been advised of
+   the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing the Work
+   or Derivative Works thereof, You may choose to offer, and charge a fee for,
+   acceptance of support, warranty, indemnity, or other liability obligations
+   and/or rights consistent with this License. However, in accepting such
+   obligations, You may act only on Your own behalf and on Your sole
+   responsibility, not on behalf of any other Contributor, and only if You
+   agree to indemnify, defend, and hold each Contributor harmless for any
+   liability incurred by, or claims asserted against, such Contributor by
+   reason of your accepting any such warranty or additional liability. END OF
+   TERMS AND CONDITIONS
+
+APPENDIX: How to apply the Apache License to your work.
+
+To apply the Apache License to your work, attach the following boilerplate
+notice, with the fields enclosed by brackets "[]" replaced with your own
+identifying information. (Don't include the brackets!) The text should be
+enclosed in the appropriate comment syntax for the file format. We also
+recommend that a file or class name and description of purpose be included on
+the same "printed page" as the copyright notice for easier identification
+within third-party archives.
+
+Copyright [yyyy] [name of copyright owner]
+
+Licensed under the Apache License, Version 2.0 (the "License");
+
+you may not use this file except in compliance with the License.
+
+You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+
+distributed under the License is distributed on an "AS IS" BASIS,
+
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+
+See the License for the specific language governing permissions and
+
+limitations under the License.
+
+* For botocore see also this required NOTICE:
+    Botocore
+    Copyright 2012-2017 Amazon.com, Inc. or its affiliates. All Rights
+    Reserved.
+
+    ----
+
+    Botocore includes a vendorized copy of the requests python library to ease
+    installation.
+
+    Requests License
+    ================
+
+    Copyright 2013 Kenneth Reitz
+
+       Licensed under the Apache License, Version 2.0 (the "License");
+       you may not use this file except in compliance with the License.
+       You may obtain a copy of the License at
+
+           http://www.apache.org/licenses/LICENSE-2.0
+
+       Unless required by applicable law or agreed to in writing, software
+       distributed under the License is distributed on an "AS IS" BASIS,
+       WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+       See the License for the specific language governing permissions and
+       limitations under the License.
+
+
+    The requests library also includes some vendorized python libraries to ease
+    installation.
+
+    Urllib3 License
+    ===============
+
+    This is the MIT license: http://www.opensource.org/licenses/mit-license.php
+
+    Copyright 2008-2011 Andrey Petrov and contributors (see CONTRIBUTORS.txt),
+    Modifications copyright 2012 Kenneth Reitz.
+
+    Permission is hereby granted, free of charge, to any person obtaining a
+    copy of this
+    software and associated documentation files (the "Software"), to deal in
+    the Software
+    without restriction, including without limitation the rights to use, copy,
+    modify, merge,
+    publish, distribute, sublicense, and/or sell copies of the Software, and to
+    permit persons
+    to whom the Software is furnished to do so, subject to the following
+    conditions:
+
+    The above copyright notice and this permission notice shall be included in
+    all copies or
+    substantial portions of the Software.
+
+    THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+    IMPLIED,
+    INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR
+    A PARTICULAR
+    PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
+    HOLDERS BE LIABLE
+    FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF
+    CONTRACT, TORT OR
+    OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+    USE OR OTHER
+    DEALINGS IN THE SOFTWARE.
+
+    Chardet License
+    ===============
+
+    This library is free software; you can redistribute it and/or
+    modify it under the terms of the GNU Lesser General Public
+    License as published by the Free Software Foundation; either
+    version 2.1 of the License, or (at your option) any later version.
+
+    This library is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+    Lesser General Public License for more details.
+
+    You should have received a copy of the GNU Lesser General Public
+    License along with this library; if not, write to the Free Software
+    Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
+    02110-1301  USA
+
+    Bundle of CA Root Certificates
+    ==============================
+
+    This library is free software; you can redistribute it and/or
+    modify it under the terms of the GNU Lesser General Public
+    License as published by the Free Software Foundation; either
+    version 2.1 of the License, or (at your option) any later version.
+
+    This library is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+    Lesser General Public License for more details.
+
+    You should have received a copy of the GNU Lesser General Public
+    License along with this library; if not, write to the Free Software
+    Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
+    02110-1301
+* For s3transfer see also this required NOTICE:
+    s3transfer
+    Copyright 2016 Amazon.com, Inc. or its affiliates. All Rights Reserved.
+
+------
+
+** colorama; version 0.4.2 -- https://pypi.org/project/colorama/
+Copyright (c) 2010 Jonathan Hartley
+All rights reserved.
+
+Copyright (c) 2010 Jonathan Hartley
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+* Redistributions of source code must retain the above copyright notice, this
+  list of conditions and the following disclaimer.
+
+* Redistributions in binary form must reproduce the above copyright notice,
+  this list of conditions and the following disclaimer in the documentation
+  and/or other materials provided with the distribution.
+
+* Neither the name of the copyright holders, nor those of its contributors
+  may be used to endorse or promote products derived from this software without
+  specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+------
+
+** prompt-toolkit; version 2.0.10 --
+https://github.com/prompt-toolkit/python-prompt-toolkit/tree/2.0.10
+Copyright (c) 2014, Jonathan Slenders
+All rights reserved.
+
+Copyright (c) 2014, Jonathan Slenders
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification,
+are permitted provided that the following conditions are met:
+
+* Redistributions of source code must retain the above copyright notice, this
+  list of conditions and the following disclaimer.
+
+* Redistributions in binary form must reproduce the above copyright notice,
+this
+  list of conditions and the following disclaimer in the documentation and/or
+  other materials provided with the distribution.
+
+* Neither the name of the {organization} nor the names of its
+  contributors may be used to endorse or promote products derived from
+  this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+------
+
+** dateutil; version 2.8.0 -- https://github.com/dateutil/dateutil/tree/2.8.0
+Copyright 2017- Paul Ganssle <[email protected]>
+Copyright 2017- dateutil contributors (see AUTHORS file)
+
+Copyright 2017- Paul Ganssle <[email protected]>
+Copyright 2017- dateutil contributors (see AUTHORS file)
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.
+
+The above license applies to all contributions after 2017-12-01, as well as
+all contributions that have been re-licensed (see AUTHORS file for the list of
+contributors who have re-licensed their code).
+--------------------------------------------------------------------------------
+dateutil - Extensions to the standard Python datetime module.
+
+Copyright (c) 2003-2011 - Gustavo Niemeyer <[email protected]>
+Copyright (c) 2012-2014 - Tomi Pieviläinen <[email protected]>
+Copyright (c) 2014-2016 - Yaron de Leeuw <[email protected]>
+Copyright (c) 2015-     - Paul Ganssle <[email protected]>
+Copyright (c) 2015-     - dateutil contributors (see AUTHORS file)
+
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+      this list of conditions and the following disclaimer.
+    * Redistributions in binary form must reproduce the above copyright notice,
+      this list of conditions and the following disclaimer in the documentation
+      and/or other materials provided with the distribution.
+    * Neither the name of the copyright holder nor the names of its
+      contributors may be used to endorse or promote products derived from
+      this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
+CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+The above BSD License Applies to all code, even that also covered by Apache
+2.0.
+
+------
+
+** Pyintaller 3.5; version 3.5 --
+https://github.com/pyinstaller/pyinstaller/tree/v3.5
+Copyright (c) 2010-2020, PyInstaller Development Team
+Copyright (c) 2005-2009, Giovanni Bajo
+Based on previous work under copyright (c) 2002 McMillan Enterprises, Inc.
+
+    * Package Pyintaller 3.5's source code may be found at:
+      https://files.pythonhosted.org/packages/e2/c9/0b44b2ea87ba36395483a672fddd07e6a9cb2b8d3c4a28d7ae76c7e7e1e5/PyInstaller-3.5.tar.gz
+
+================================
+ The PyInstaller licensing terms
+================================
+
+
+Copyright (c) 2010-2020, PyInstaller Development Team
+Copyright (c) 2005-2009, Giovanni Bajo
+Based on previous work under copyright (c) 2002 McMillan Enterprises, Inc.
+
+
+PyInstaller is licensed under the terms of the GNU General Public License
+as published by the Free Software Foundation; either version 2 of the License,
+or (at your option) any later version.
+
+
+Bootloader Exception
+--------------------
+
+In addition to the permissions in the GNU General Public License, the
+authors give you unlimited permission to link or embed compiled bootloader
+and related files into combinations with other programs, and to distribute
+those combinations without any restriction coming from the use of those
+files. (The General Public License restrictions do apply in other respects;
+for example, they cover modification of the files, and distribution when
+not linked into a combined executable.)
+
+
+Bootloader and Related Files
+----------------------------
+
+Bootloader and related files are files which are embedded within the
+final executable. This includes files in directories:
+
+./bootloader/
+./PyInstaller/loader
+
+
+About the PyInstaller Development Team
+--------------------------------------
+
+The PyInstaller Development Team is the set of contributors
+to the PyInstaller project. A full list with details is kept
+in the documentation directory, in the file
+``doc/CREDITS.rst``.
+
+The core team that coordinates development on GitHub can be found here:
+https://github.com/pyinstaller/pyinstaller.  As of 2015, it consists of:
+
+* Hartmut Goebel
+* Martin Zibricky
+* David Vierra
+* David Cortesi
+
+
+Our Copyright Policy
+--------------------
+
+PyInstaller uses a shared copyright model. Each contributor maintains copyright
+over their contributions to PyInstaller. But, it is important to note that
+these
+contributions are typically only changes to the repositories. Thus,
+the PyInstaller source code, in its entirety is not the copyright of any single
+person or institution.  Instead, it is the collective copyright of the entire
+PyInstaller Development Team.  If individual contributors want to maintain
+a record of what changes/contributions they have specific copyright on, they
+should indicate their copyright in the commit message of the change, when they
+commit the change to the PyInstaller repository.
+
+With this in mind, the following banner should be used in any source code file
+to indicate the copyright and license terms:
+
+
+#-----------------------------------------------------------------------------
+# Copyright (c) 2005-2020, PyInstaller Development Team.
+#
+# Distributed under the terms of the GNU General Public License (version 2
+# or later) with exception for distributing the bootloader.
+#
+# The full license is in the file COPYING.txt, distributed with this software.
+#
+# SPDX-License-Identifier: (GPL-2.0-or-later WITH Bootloader-exception)
+#-----------------------------------------------------------------------------
+
+
+
+GNU General Public License
+--------------------------
+
+https://gnu.org/licenses/gpl-2.0.html
+
+
+		    GNU GENERAL PUBLIC LICENSE
+		       Version 2, June 1991
+
+ Copyright (C) 1989, 1991 Free Software Foundation, Inc.
+ 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+			    Preamble
+
+  The licenses for most software are designed to take away your
+freedom to share and change it.  By contrast, the GNU General Public
+License is intended to guarantee your freedom to share and change free
+software--to make sure the software is free for all its users.  This
+General Public License applies to most of the Free Software
+Foundation's software and to any other program whose authors commit to
+using it.  (Some other Free Software Foundation software is covered by
+the GNU Library General Public License instead.)  You can apply it to
+your programs, too.
+
+  When we speak of free software, we are referring to freedom, not
+price.  Our General Public Licenses are designed to make sure that you
+have the freedom to distribute copies of free software (and charge for
+this service if you wish), that you receive source code or can get it
+if you want it, that you can change the software or use pieces of it
+in new free programs; and that you know you can do these things.
+
+  To protect your rights, we need to make restrictions that forbid
+anyone to deny you these rights or to ask you to surrender the rights.
+These restrictions translate to certain responsibilities for you if you
+distribute copies of the software, or if you modify it.
+
+  For example, if you distribute copies of such a program, whether
+gratis or for a fee, you must give the recipients all the rights that
+you have.  You must make sure that they, too, receive or can get the
+source code.  And you must show them these terms so they know their
+rights.
+
+  We protect your rights with two steps: (1) copyright the software, and
+(2) offer you this license which gives you legal permission to copy,
+distribute and/or modify the software.
+
+  Also, for each author's protection and ours, we want to make certain
+that everyone understands that there is no warranty for this free
+software.  If the software is modified by someone else and passed on, we
+want its recipients to know that what they have is not the original, so
+that any problems introduced by others will not reflect on the original
+authors' reputations.
+
+  Finally, any free program is threatened constantly by software
+patents.  We wish to avoid the danger that redistributors of a free
+program will individually obtain patent licenses, in effect making the
+program proprietary.  To prevent this, we have made it clear that any
+patent must be licensed for everyone's free use or not licensed at all.
+
+  The precise terms and conditions for copying, distribution and
+modification follow.
+
+		    GNU GENERAL PUBLIC LICENSE
+   TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
+
+  0. This License applies to any program or other work which contains
+a notice placed by the copyright holder saying it may be distributed
+under the terms of this General Public License.  The "Program", below,
+refers to any such program or work, and a "work based on the Program"
+means either the Program or any derivative work under copyright law:
+that is to say, a work containing the Program or a portion of it,
+either verbatim or with modifications and/or translated into another
+language.  (Hereinafter, translation is included without limitation in
+the term "modification".)  Each licensee is addressed as "you".
+
+Activities other than copying, distribution and modification are not
+covered by this License; they are outside its scope.  The act of
+running the Program is not restricted, and the output from the Program
+is covered only if its contents constitute a work based on the
+Program (independent of having been made by running the Program).
+Whether that is true depends on what the Program does.
+
+  1. You may copy and distribute verbatim copies of the Program's
+source code as you receive it, in any medium, provided that you
+conspicuously and appropriately publish on each copy an appropriate
+copyright notice and disclaimer of warranty; keep intact all the
+notices that refer to this License and to the absence of any warranty;
+and give any other recipients of the Program a copy of this License
+along with the Program.
+
+You may charge a fee for the physical act of transferring a copy, and
+you may at your option offer warranty protection in exchange for a fee.
+
+  2. You may modify your copy or copies of the Program or any portion
+of it, thus forming a work based on the Program, and copy and
+distribute such modifications or work under the terms of Section 1
+above, provided that you also meet all of these conditions:
+
+    a) You must cause the modified files to carry prominent notices
+    stating that you changed the files and the date of any change.
+
+    b) You must cause any work that you distribute or publish, that in
+    whole or in part contains or is derived from the Program or any
+    part thereof, to be licensed as a whole at no charge to all third
+    parties under the terms of this License.
+
+    c) If the modified program normally reads commands interactively
+    when run, you must cause it, when started running for such
+    interactive use in the most ordinary way, to print or display an
+    announcement including an appropriate copyright notice and a
+    notice that there is no warranty (or else, saying that you provide
+    a warranty) and that users may redistribute the program under
+    these conditions, and telling the user how to view a copy of this
+    License.  (Exception: if the Program itself is interactive but
+    does not normally print such an announcement, your work based on
+    the Program is not required to print an announcement.)
+
+These requirements apply to the modified work as a whole.  If
+identifiable sections of that work are not derived from the Program,
+and can be reasonably considered independent and separate works in
+themselves, then this License, and its terms, do not apply to those
+sections when you distribute them as separate works.  But when you
+distribute the same sections as part of a whole which is a work based
+on the Program, the distribution of the whole must be on the terms of
+this License, whose permissions for other licensees extend to the
+entire whole, and thus to each and every part regardless of who wrote it.
+
+Thus, it is not the intent of this section to claim rights or contest
+your rights to work written entirely by you; rather, the intent is to
+exercise the right to control the distribution of derivative or
+collective works based on the Program.
+
+In addition, mere aggregation of another work not based on the Program
+with the Program (or with a work based on the Program) on a volume of
+a storage or distribution medium does not bring the other work under
+the scope of this License.
+
+  3. You may copy and distribute the Program (or a work based on it,
+under Section 2) in object code or executable form under the terms of
+Sections 1 and 2 above provided that you also do one of the following:
+
+    a) Accompany it with the complete corresponding machine-readable
+    source code, which must be distributed under the terms of Sections
+    1 and 2 above on a medium customarily used for software interchange; or,
+
+    b) Accompany it with a written offer, valid for at least three
+    years, to give any third party, for a charge no more than your
+    cost of physically performing source distribution, a complete
+    machine-readable copy of the corresponding source code, to be
+    distributed under the terms of Sections 1 and 2 above on a medium
+    customarily used for software interchange; or,
+
+    c) Accompany it with the information you received as to the offer
+    to distribute corresponding source code.  (This alternative is
+    allowed only for noncommercial distribution and only if you
+    received the program in object code or executable form with such
+    an offer, in accord with Subsection b above.)
+
+The source code for a work means the preferred form of the work for
+making modifications to it.  For an executable work, complete source
+code means all the source code for all modules it contains, plus any
+associated interface definition files, plus the scripts used to
+control compilation and installation of the executable.  However, as a
+special exception, the source code distributed need not include
+anything that is normally distributed (in either source or binary
+form) with the major components (compiler, kernel, and so on) of the
+operating system on which the executable runs, unless that component
+itself accompanies the executable.
+
+If distribution of executable or object code is made by offering
+access to copy from a designated place, then offering equivalent
+access to copy the source code from the same place counts as
+distribution of the source code, even though third parties are not
+compelled to copy the source along with the object code.
+
+  4. You may not copy, modify, sublicense, or distribute the Program
+except as expressly provided under this License.  Any attempt
+otherwise to copy, modify, sublicense or distribute the Program is
+void, and will automatically terminate your rights under this License.
+However, parties who have received copies, or rights, from you under
+this License will not have their licenses terminated so long as such
+parties remain in full compliance.
+
+  5. You are not required to accept this License, since you have not
+signed it.  However, nothing else grants you permission to modify or
+distribute the Program or its derivative works.  These actions are
+prohibited by law if you do not accept this License.  Therefore, by
+modifying or distributing the Program (or any work based on the
+Program), you indicate your acceptance of this License to do so, and
+all its terms and conditions for copying, distributing or modifying
+the Program or works based on it.
+
+  6. Each time you redistribute the Program (or any work based on the
+Program), the recipient automatically receives a license from the
+original licensor to copy, distribute or modify the Program subject to
+these terms and conditions.  You may not impose any further
+restrictions on the recipients' exercise of the rights granted herein.
+You are not responsible for enforcing compliance by third parties to
+this License.
+
+  7. If, as a consequence of a court judgment or allegation of patent
+infringement or for any other reason (not limited to patent issues),
+conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License.  If you cannot
+distribute so as to satisfy simultaneously your obligations under this
+License and any other pertinent obligations, then as a consequence you
+may not distribute the Program at all.  For example, if a patent
+license would not permit royalty-free redistribution of the Program by
+all those who receive copies directly or indirectly through you, then
+the only way you could satisfy both it and this License would be to
+refrain entirely from distribution of the Program.
+
+If any portion of this section is held invalid or unenforceable under
+any particular circumstance, the balance of the section is intended to
+apply and the section as a whole is intended to apply in other
+circumstances.
+
+It is not the purpose of this section to induce you to infringe any
+patents or other property right claims or to contest validity of any
+such claims; this section has the sole purpose of protecting the
+integrity of the free software distribution system, which is
+implemented by public license practices.  Many people have made
+generous contributions to the wide range of software distributed
+through that system in reliance on consistent application of that
+system; it is up to the author/donor to decide if he or she is willing
+to distribute software through any other system and a licensee cannot
+impose that choice.
+
+This section is intended to make thoroughly clear what is believed to
+be a consequence of the rest of this License.
+
+  8. If the distribution and/or use of the Program is restricted in
+certain countries either by patents or by copyrighted interfaces, the
+original copyright holder who places the Program under this License
+may add an explicit geographical distribution limitation excluding
+those countries, so that distribution is permitted only in or among
+countries not thus excluded.  In such case, this License incorporates
+the limitation as if written in the body of this License.
+
+  9. The Free Software Foundation may publish revised and/or new versions
+of the General Public License from time to time.  Such new versions will
+be similar in spirit to the present version, but may differ in detail to
+address new problems or concerns.
+
+Each version is given a distinguishing version number.  If the Program
+specifies a version number of this License which applies to it and "any
+later version", you have the option of following the terms and conditions
+either of that version or of any later version published by the Free
+Software Foundation.  If the Program does not specify a version number of
+this License, you may choose any version ever published by the Free Software
+Foundation.
+
+  10. If you wish to incorporate parts of the Program into other free
+programs whose distribution conditions are different, write to the author
+to ask for permission.  For software which is copyrighted by the Free
+Software Foundation, write to the Free Software Foundation; we sometimes
+make exceptions for this.  Our decision will be guided by the two goals
+of preserving the free status of all derivatives of our free software and
+of promoting the sharing and reuse of software generally.
+
+			    NO WARRANTY
+
+  11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
+FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW.  EXCEPT WHEN
+OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
+PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
+OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.  THE ENTIRE RISK AS
+TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU.  SHOULD THE
+PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
+REPAIR OR CORRECTION.
+
+  12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
+REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
+INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
+OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
+TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
+YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
+PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
+POSSIBILITY OF SUCH DAMAGES.
+
+		     END OF TERMS AND CONDITIONS
+
+------
+
+** six; version 1.14.0 -- https://github.com/benjaminp/six/tree/1.14.0
+Copyright (c) 2010-2020 Benjamin Peterson
+
+Copyright (c) 2010-2020 Benjamin Peterson
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of
+this software and associated documentation files (the "Software"), to deal in
+the Software without restriction, including without limitation the rights to
+use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
+of
+the Software, and to permit persons to whom the Software is furnished to do so,
+subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS
+FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
+COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
+IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+------
+
+** urllib3; version 1.25.8 -- https://github.com/urllib3/urllib3/tree/1.25.8
+Copyright (c) 2008-2019 Andrey Petrov and contributors (see CONTRIBUTORS.txt)
+
+MIT License
+
+Copyright (c) 2008-2019 Andrey Petrov and contributors (see CONTRIBUTORS.txt)
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+
+------
+
+** setuptools; version 45.2.0 --
+https://github.com/pypa/setuptools/tree/v45.2.0
+Copyright (C) 2016 Jason R Coombs <[email protected]>
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of
+this software and associated documentation files (the "Software"), to deal in
+the Software without restriction, including without limitation the rights to
+use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
+of the Software, and to permit persons to whom the Software is furnished to do
+so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+
+------
+
+** wcwidth; version 0.1.8 -- https://github.com/jquast/wcwidth/tree/0.1.8
+Copyright (c) 2014 Jeff Quast <[email protected]>
+
+The MIT License (MIT)
+
+Copyright (c) 2014 Jeff Quast <[email protected]>
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+
+------
+
+** cffi; version 1.14.0 --
+https://foss.heptapod.net/pypy/cffi/tree/branch/release-1.14
+© Copyright 2012-2018, Armin Rigo, Maciej Fijalkowski
+
+Except when otherwise stated (look for LICENSE files in directories or
+information at the beginning of each file) all software and
+documentation is licensed as follows:
+
+    The MIT License
+
+    Permission is hereby granted, free of charge, to any person
+    obtaining a copy of this software and associated documentation
+    files (the "Software"), to deal in the Software without
+    restriction, including without limitation the rights to use,
+    copy, modify, merge, publish, distribute, sublicense, and/or
+    sell copies of the Software, and to permit persons to whom the
+    Software is furnished to do so, subject to the following conditions:
+
+    The above copyright notice and this permission notice shall be included
+    in all copies or substantial portions of the Software.
+
+    THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+    OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+    FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+    THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+    LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+    FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+    DEALINGS IN THE SOFTWARE.
+
+------
+
+** jmespath; version 0.9.4 --
+https://github.com/jmespath/jmespath.py/tree/0.9.4
+Copyright (c) 2013 Amazon.com, Inc. or its affiliates.  All Rights Reserved
+
+Copyright (c) 2013 Amazon.com, Inc. or its affiliates.  All Rights Reserved
+
+Permission is hereby granted, free of charge, to any person obtaining a
+copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish, dis-
+tribute, sublicense, and/or sell copies of the Software, and to permit
+persons to whom the Software is furnished to do so, subject to the fol-
+lowing conditions:
+
+The above copyright notice and this permission notice shall be included
+in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABIL-
+ITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT
+SHALL THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+IN THE SOFTWARE.
+
+------
+
+** ruamel.yaml; version 0.15.100 --
+https://sourceforge.net/p/ruamel-yaml/code/ci/default/tree/
+Copyright (c) 2014-2019 Anthon van der Neut, Ruamel bvba
+
+ The MIT License (MIT)
+
+ Copyright (c) 2014-2020 Anthon van der Neut, Ruamel bvba
+
+ Permission is hereby granted, free of charge, to any person obtaining a copy
+ of this software and associated documentation files (the "Software"), to deal
+ in the Software without restriction, including without limitation the rights
+ to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ copies of the Software, and to permit persons to whom the Software is
+ furnished to do so, subject to the following conditions:
+
+ The above copyright notice and this permission notice shall be included in
+ all copies or substantial portions of the Software.
+
+ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ SOFTWARE.
+
+------
+
+** OpenSSL; version 1.0.2s --
+https://github.com/openssl/openssl/tree/OpenSSL_1_0_1s
+Copyright (c) 1998-2011 The OpenSSL Project.  All rights reserved.
+Copyright (C) 1995-1998 Eric Young ([email protected])
+
+LICENSE ISSUES
+  ==============
+
+  The OpenSSL toolkit stays under a double license, i.e. both the conditions of
+  the OpenSSL License and the original SSLeay license apply to the toolkit.
+  See below for the actual license texts.
+
+  OpenSSL License
+  ---------------
+
+/* ====================================================================
+ * Copyright (c) 1998-2017 The OpenSSL Project.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *
+ * 3. All advertising materials mentioning features or use of this
+ *    software must display the following acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
+ *
+ * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
+ *    endorse or promote products derived from this software without
+ *    prior written permission. For written permission, please contact
+ *    [email protected].
+ *
+ * 5. Products derived from this software may not be called "OpenSSL"
+ *    nor may "OpenSSL" appear in their names without prior written
+ *    permission of the OpenSSL Project.
+ *
+ * 6. Redistributions of any form whatsoever must retain the following
+ *    acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit (http://www.openssl.org/)"
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
+ * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OpenSSL PROJECT OR
+ * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ * OF THE POSSIBILITY OF SUCH DAMAGE.
+ * ====================================================================
+ *
+ * This product includes cryptographic software written by Eric Young
+ * ([email protected]).  This product includes software written by Tim
+ * Hudson ([email protected]).
+ *
+ */
+
+ Original SSLeay License
+ -----------------------
+
+/* Copyright (C) 1995-1998 Eric Young ([email protected])
+ * All rights reserved.
+ *
+ * This package is an SSL implementation written
+ * by Eric Young ([email protected]).
+ * The implementation was written so as to conform with Netscapes SSL.
+ *
+ * This library is free for commercial and non-commercial use as long as
+ * the following conditions are aheared to.  The following conditions
+ * apply to all code found in this distribution, be it the RC4, RSA,
+ * lhash, DES, etc., code; not just the SSL code.  The SSL documentation
+ * included with this distribution is covered by the same copyright terms
+ * except that the holder is Tim Hudson ([email protected]).
+ *
+ * Copyright remains Eric Young's, and as such any Copyright notices in
+ * the code are not to be removed.
+ * If this package is used in a product, Eric Young should be given attribution
+ * as the author of the parts of the library used.
+ * This can be in the form of a textual message at program startup or
+ * in documentation (online or textual) provided with the package.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. All advertising materials mentioning features or use of this software
+ *    must display the following acknowledgement:
+ *    "This product includes cryptographic software written by
+ *     Eric Young ([email protected])"
+ *    The word 'cryptographic' can be left out if the rouines from the library
+ *    being used are not cryptographic related :-).
+ * 4. If you include any Windows specific code (or a derivative thereof) from
+ * the apps directory (application code) you must include an acknowledgement:
+ * "This product includes software written by Tim Hudson ([email protected])"
+ *
+ * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * The licence and distribution terms for any publically available version or
+ * derivative of this code cannot be changed.  i.e. this code cannot simply be
+ * copied and put under another distribution licence
+ * [including the GNU Public Licence.]
+ */
+
+------
+
+** Python 3.9.11; version 3.9.11 -- https://github.com/python/cpython/tree/v3.9.11
+Copyright © 2001-2020 Python Software Foundation. All rights reserved.
+
+PYTHON SOFTWARE FOUNDATION LICENSE VERSION 2
+--------------------------------------------
+
+1. This LICENSE AGREEMENT is between the Python Software Foundation
+("PSF"), and the Individual or Organization ("Licensee") accessing and
+otherwise using this software ("Python") in source or binary form and
+its associated documentation.
+
+2. Subject to the terms and conditions of this License Agreement, PSF hereby
+grants Licensee a nonexclusive, royalty-free, world-wide license to reproduce,
+analyze, test, perform and/or display publicly, prepare derivative works,
+distribute, and otherwise use Python alone or in any derivative version,
+provided, however, that PSF's License Agreement and PSF's notice of copyright,
+i.e., "Copyright (c) 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009,
+2010,
+2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020 Python Software
+Foundation;
+All Rights Reserved" are retained in Python alone or in any derivative version
+prepared by Licensee.
+
+3. In the event Licensee prepares a derivative work that is based on
+or incorporates Python or any part thereof, and wants to make
+the derivative work available to others as provided herein, then
+Licensee hereby agrees to include in any such work a brief summary of
+the changes made to Python.
+
+4. PSF is making Python available to Licensee on an "AS IS"
+basis.  PSF MAKES NO REPRESENTATIONS OR WARRANTIES, EXPRESS OR
+IMPLIED.  BY WAY OF EXAMPLE, BUT NOT LIMITATION, PSF MAKES NO AND
+DISCLAIMS ANY REPRESENTATION OR WARRANTY OF MERCHANTABILITY OR FITNESS
+FOR ANY PARTICULAR PURPOSE OR THAT THE USE OF PYTHON WILL NOT
+INFRINGE ANY THIRD PARTY RIGHTS.
+
+5. PSF SHALL NOT BE LIABLE TO LICENSEE OR ANY OTHER USERS OF PYTHON
+FOR ANY INCIDENTAL, SPECIAL, OR CONSEQUENTIAL DAMAGES OR LOSS AS
+A RESULT OF MODIFYING, DISTRIBUTING, OR OTHERWISE USING PYTHON,
+OR ANY DERIVATIVE THEREOF, EVEN IF ADVISED OF THE POSSIBILITY THEREOF.
+
+6. This License Agreement will automatically terminate upon a material
+breach of its terms and conditions.
+
+7. Nothing in this License Agreement shall be deemed to create any
+relationship of agency, partnership, or joint venture between PSF and
+Licensee.  This License Agreement does not grant permission to use PSF
+trademarks or trade name in a trademark sense to endorse or promote
+products or services of Licensee, or any third party.
+
+8. By copying, installing or otherwise using Python, Licensee
+agrees to be bound by the terms and conditions of this License
+Agreement.
+
+
+
+------
+
+** docutils; version 0.15.2 --
+https://sourceforge.net/p/docutils/code/HEAD/tree/trunk/docutils/
+:Author: David Goodger
+:Contact: [email protected]
+:Date: $Date: 2015-05-08 15:56:32 +0000 (Fri, 08 May 2015) $
+:Web site: http://docutils.sourceforge.net/
+:Copyright: This document has been placed in the public domain.
+
+==================
+ Copying Docutils
+==================
+
+:Author: David Goodger
+:Contact: [email protected]
+:Date: $Date: 2015-05-08 15:56:32 +0000 (Fri, 08 May 2015) $
+:Web site: http://docutils.sourceforge.net/
+:Copyright: This document has been placed in the public domain.
+
+Most of the files included in this project have been placed in the
+public domain, and therefore have no license requirements and no
+restrictions on copying or usage; see the `Public Domain Dedication`_
+below.  There are a few exceptions_, listed below.
+Files in the Sandbox_ are not distributed with Docutils releases and
+may have different license terms.
+
+
+Public Domain Dedication
+========================
+
+The persons who have associated their work with this project (the
+"Dedicator": David Goodger and the many contributors to the Docutils
+project) hereby dedicate the entire copyright, less the exceptions_
+listed below, in the work of authorship known as "Docutils" identified
+below (the "Work") to the public domain.
+
+The primary repository for the Work is the Internet World Wide Web
+site <http://docutils.sourceforge.net/>.  The Work consists of the
+files within the "docutils" module of the Docutils project Subversion
+repository (Internet host docutils.svn.sourceforge.net, filesystem path
+/svnroot/docutils), whose Internet web interface is located at
+<http://docutils.svn.sourceforge.net/viewvc/docutils/>.  Files dedicated to the
+public domain may be identified by the inclusion, near the beginning
+of each file, of a declaration of the form::
+
+    Copyright: This document/module/DTD/stylesheet/file/etc. has been
+               placed in the public domain.
+
+Dedicator makes this dedication for the benefit of the public at large
+and to the detriment of Dedicator's heirs and successors.  Dedicator
+intends this dedication to be an overt act of relinquishment in
+perpetuity of all present and future rights under copyright law,
+whether vested or contingent, in the Work.  Dedicator understands that
+such relinquishment of all rights includes the relinquishment of all
+rights to enforce (by lawsuit or otherwise) those copyrights in the
+Work.
+
+Dedicator recognizes that, once placed in the public domain, the Work
+may be freely reproduced, distributed, transmitted, used, modified,
+built upon, or otherwise exploited by anyone for any purpose,
+commercial or non-commercial, and in any way, including by methods
+that have not yet been invented or conceived.
+
+(This dedication is derived from the text of the `Creative Commons
+Public Domain Dedication`. [#]_)
+
+.. [#] Creative Commons has `retired this legal tool`__ and does not
+   recommend that it be applied to works: This tool is based on United
+   States law and may not be applicable outside the US. For dedicating new
+   works to the public domain, Creative Commons recommend the replacement
+   Public Domain Dedication CC0_ (CC zero, "No Rights Reserved"). So does
+   the Free Software Foundation in its license-list_.
+
+   __  http://creativecommons.org/retiredlicenses
+   .. _CC0: http://creativecommons.org/about/cc0
+
+Exceptions
+==========
+
+The exceptions to the `Public Domain Dedication`_ above are:
+
+* docutils/writers/s5_html/themes/default/iepngfix.htc:
+
+      IE5.5+ PNG Alpha Fix v1.0 by Angus Turnbull
+      <http://www.twinhelix.com>.  Free usage permitted as long as
+      this notice remains intact.
+
+* docutils/utils/math/__init__.py,
+  docutils/utils/math/latex2mathml.py,
+  docutils/writers/xetex/__init__.py,
+  docutils/writers/latex2e/docutils-05-compat.sty,
+  docs/user/docutils-05-compat.sty.txt,
+  docutils/utils/error_reporting.py,
+  docutils/test/transforms/test_smartquotes.py:
+
+  Copyright © Günter Milde.
+  Released under the terms of the `2-Clause BSD license`_
+  (`local copy <licenses/BSD-2-Clause.txt>`__).
+
+* docutils/utils/smartquotes.py
+
+  Copyright © 2011 Günter Milde,
+  based on `SmartyPants`_ © 2003 John Gruber
+  (released under a 3-Clause BSD license included in the file)
+  and smartypants.py © 2004, 2007 Chad Miller.
+  Released under the terms of the `2-Clause BSD license`_
+  (`local copy <licenses/BSD-2-Clause.txt>`__).
+
+  .. _SmartyPants: http://daringfireball.net/projects/smartypants/
+
+* docutils/utils/math/math2html.py,
+  docutils/writers/html4css1/math.css
+
+  Copyright © Alex Fernández
+  These files are part of eLyXer_, released under the `GNU
+  General Public License`_ version 3 or later. The author relicensed
+  them for Docutils under the terms of the `2-Clause BSD license`_
+  (`local copy <licenses/BSD-2-Clause.txt>`__).
+
+  .. _eLyXer: http://www.nongnu.org/elyxer/
+
+* docutils/utils/roman.py, copyright by Mark Pilgrim, released under the
+  `Python 2.1.1 license`_ (`local copy`__).
+
+  __ licenses/python-2-1-1.txt
+
+* tools/editors/emacs/rst.el, copyright by Free Software Foundation,
+  Inc., released under the `GNU General Public License`_ version 3 or
+  later (`local copy`__).
+
+  __ licenses/gpl-3-0.txt
+
+The `2-Clause BSD license`_ and the Python licenses are OSI-approved_
+and GPL-compatible_.
+
+Plaintext versions of all the linked-to licenses are provided in the
+licenses_ directory.
+
+.. _sandbox: http://docutils.sourceforge.net/sandbox/README.html
+.. _licenses: licenses/
+.. _Python 2.1.1 license: http://www.python.org/2.1.1/license.html
+.. _GNU General Public License: http://www.gnu.org/copyleft/gpl.html
+.. _2-Clause BSD license: http://www.spdx.org/licenses/BSD-2-Clause
+.. _OSI-approved: http://opensource.org/licenses/
+.. _license-list:
+.. _GPL-compatible: http://www.gnu.org/licenses/license-list.html

aws/install

@@ -0,0 +1,155 @@
+#!/bin/sh
+# Copyright 2012-2019 Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+
+usage() {
+  cat 1>&2 <<EOF
+Installs the AWS CLI v2
+
+USAGE:
+    install [FLAGS] [OPTIONS]
+
+FLAGS:
+    -u, --update              Updates the AWS CLI v2 if a different version
+                              is previously installed. By default, this script
+                              will not update the AWS CLI if a previous
+                              installation is detected.
+
+    -h, --help                Prints help information
+
+OPTIONS:
+    -i, --install-dir <path>  The directory to install the AWS CLI v2. By
+                              default, this directory is: /usr/local/aws-cli
+
+    -b, --bin-dir <path>      The directory to store symlinks to executables
+                              for the AWS CLI v2. By default, the directory
+                              used is: /usr/local/bin
+EOF
+}
+
+parse_commandline() {
+  while test $# -gt 0
+  do
+    key="$1"
+	case "$key" in
+	  -i|--install-dir)
+	    PARSED_INSTALL_DIR="$2"
+		shift
+	   ;;
+	  -b|--bin-dir)
+	    PARSED_BIN_DIR="$2"
+		shift
+	   ;;
+	  -u|--update)
+	    PARSED_UPGRADE="yes"
+	  ;;
+	  -h|--help)
+	    usage
+        exit 0
+	  ;;
+	  *)
+	   die "Got an unexpected argument: $1"
+	  ;;
+    esac
+	shift
+  done
+}
+
+set_global_vars() {
+  ROOT_INSTALL_DIR=${PARSED_INSTALL_DIR:-/usr/local/aws-cli}
+  BIN_DIR=${PARSED_BIN_DIR:-/usr/local/bin}
+  UPGRADE=${PARSED_UPGRADE:-no}
+
+  EXE_NAME="aws"
+  COMPLETER_EXE_NAME="aws_completer"
+  INSTALLER_DIR="$( cd "$( dirname "$0" )" >/dev/null 2>&1 && pwd )"
+  INSTALLER_DIST_DIR="$INSTALLER_DIR/dist"
+  INSTALLER_EXE="$INSTALLER_DIST_DIR/$EXE_NAME"
+  AWS_EXE_VERSION=$($INSTALLER_EXE --version | cut -d ' ' -f 1 | cut -d '/' -f 2)
+
+  INSTALL_DIR="$ROOT_INSTALL_DIR/v2/$AWS_EXE_VERSION"
+  INSTALL_DIR="$INSTALL_DIR"
+  INSTALL_DIST_DIR="$INSTALL_DIR/dist"
+  INSTALL_BIN_DIR="$INSTALL_DIR/bin"
+  INSTALL_AWS_EXE="$INSTALL_BIN_DIR/$EXE_NAME"
+  INSTALL_AWS_COMPLETER_EXE="$INSTALL_BIN_DIR/$COMPLETER_EXE_NAME"
+
+  CURRENT_INSTALL_DIR="$ROOT_INSTALL_DIR/v2/current"
+  CURRENT_AWS_EXE="$CURRENT_INSTALL_DIR/bin/$EXE_NAME"
+  CURRENT_AWS_COMPLETER_EXE="$CURRENT_INSTALL_DIR/bin/$COMPLETER_EXE_NAME"
+
+  BIN_AWS_EXE="$BIN_DIR/$EXE_NAME"
+  BIN_AWS_COMPLETER_EXE="$BIN_DIR/$COMPLETER_EXE_NAME"
+}
+
+create_install_dir() {
+  mkdir -p "$INSTALL_DIR" || exit 1
+  {
+    setup_install_dist &&
+    setup_install_bin &&
+    create_current_symlink
+  } || {
+    rm -rf "$INSTALL_DIR"
+    exit 1
+  }
+}
+
+check_preexisting_install() {
+  if [ -L "$CURRENT_INSTALL_DIR" ] && [ "$UPGRADE" = "no" ]
+  then
+    die "Found preexisting AWS CLI installation: $CURRENT_INSTALL_DIR. Please rerun install script with --update flag."
+  fi
+  if [ -d "$INSTALL_DIR" ]
+  then
+    echo "Found same AWS CLI version: $INSTALL_DIR. Skipping install."
+    exit 0
+  fi
+}
+
+setup_install_dist() {
+  cp -r "$INSTALLER_DIST_DIR" "$INSTALL_DIST_DIR"
+}
+
+setup_install_bin() {
+  mkdir -p "$INSTALL_BIN_DIR"
+  ln -s "../dist/$EXE_NAME" "$INSTALL_AWS_EXE"
+  ln -s "../dist/$COMPLETER_EXE_NAME" "$INSTALL_AWS_COMPLETER_EXE"
+}
+
+create_current_symlink() {
+  ln -snf "$INSTALL_DIR" "$CURRENT_INSTALL_DIR"
+}
+
+create_bin_symlinks() {
+  mkdir -p "$BIN_DIR"
+  ln -sf "$CURRENT_AWS_EXE" "$BIN_AWS_EXE"
+  ln -sf "$CURRENT_AWS_COMPLETER_EXE" "$BIN_AWS_COMPLETER_EXE"
+}
+
+die() {
+	err_msg="$1"
+	echo "$err_msg" >&2
+	exit 1
+}
+
+main() {
+  parse_commandline "$@"
+  set_global_vars
+  check_preexisting_install
+  create_install_dir
+  create_bin_symlinks
+  echo "You can now run: $BIN_AWS_EXE --version"
+  exit 0
+}
+
+main "$@" || exit 1

awscliv2.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41ced963f9c2193fe875d7b21c5873980cc482ae7f054e5db9a7f66b11069624
+size 66108930

pyproject.toml

@@ -0,0 +1,41 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "videollama2"
+version = "1.0"
+description = "Release of VideoLLaMA2"
+readme = "README.md"
+requires-python = ">=3.8"
+classifiers = [
+    "Programming Language :: Python :: 3",
+    "License :: OSI Approved :: Apache Software License",
+]
+dependencies = [
+    "torch==2.2.0", "torchvision==0.17.0", 
+    "transformers==4.40.0", "tokenizers==0.19.1", 
+    "deepspeed==0.13.1", "accelerate==0.26.1",
+    "peft==0.4.0", "timm==1.0.3", "numpy==1.24.4",
+    "decord==0.6.0", "imageio==2.34.0", "imageio-ffmpeg==0.4.9",
+    "moviepy==1.0.3", "scenedetect==0.6.3",
+    "opencv-python==4.6.0.66", "pysubs2",
+    "scikit-learn==1.2.2", "huggingface_hub==0.23.4", "sentencepiece==0.1.99",
+    "shortuuid", "einops==0.6.1", "einops-exts==0.0.4", 
+    "bitsandbytes==0.43.0", "pydantic>=2.0", "markdown2[all]", 
+    "gradio==3.50.0", "gradio_client==0.6.1", "httpx==0.24.1",
+    "requests", "openai", "uvicorn", "fastapi", "tensorboard", "wandb", "tabulate"
+]
+
+[project.optional-dependencies]
+train = ["ninja"]
+
+[project.urls]
+"Homepage" = "https://github.com/DAMO-NLP-SG/VideoLLaMA2"
+"Bug Tracker" = "https://github.com/DAMO-NLP-SG/VideoLLaMA2/issues"
+
+[tool.setuptools.packages.find]
+exclude = ["assets*", "benchmark*", "docs", "dist*", "playground*", "scripts*", "tests*"]
+
+[tool.wheel]
+exclude = ["assets*", "benchmark*", "docs", "dist*", "playground*", "scripts*", "tests*"]

requirements.txt

@@ -0,0 +1,40 @@
+--extra-index-url https://download.pytorch.org/whl/cu118
+# basic dependencies
+torch==2.2.0
+torchvision==0.17.0
+transformers==4.40.0
+tokenizers==0.19.1
+deepspeed==0.13.1
+accelerate==0.26.1
+peft==0.4.0
+timm==1.0.3
+numpy==1.24.4
+# data processing
+decord==0.6.0
+imageio==2.34.0
+imageio-ffmpeg==0.4.9
+moviepy==1.0.3
+scenedetect==0.6.3
+opencv-python==4.6.0.66
+pysubs2
+# misc
+scikit-learn==1.2.2
+huggingface_hub==0.23.4
+sentencepiece==0.1.99
+shortuuid
+einops==0.6.1
+einops-exts==0.0.4
+bitsandbytes==0.43.0
+pydantic>=2.0
+markdown2[all]
+gradio==3.50.0
+gradio_client==0.6.1
+httpx==0.24.1
+requests
+openai
+uvicorn
+fastapi
+tensorboard
+wandb
+tabulate
+spaces==0.29.2

scripts/custom/finetune.sh

@@ -0,0 +1,74 @@
+#!/bin/bash
+
+# Environment Variables
+ARG_WORLD_SIZE=${1:-1}
+ARG_NPROC_PER_NODE=${2:-8}
+ARG_MASTER_ADDR="127.0.0.1"
+ARG_MASTER_PORT=16666
+ARG_RANK=0
+
+# Multiple conditions
+if [ ! -n "$WORLD_SIZE" ] || [ ! -n "$NPROC_PER_NODE" ]; then
+    WORLD_SIZE=$ARG_WORLD_SIZE
+    NPROC_PER_NODE=$ARG_NPROC_PER_NODE
+fi
+if [ ! -n "$MASTER_ADDR" ] || [ ! -n "$MASTER_PORT" ] || [ ! -n "$RANK" ]; then
+    MASTER_ADDR=$ARG_MASTER_ADDR
+    MASTER_PORT=$ARG_MASTER_PORT
+    RANK=$ARG_RANK
+fi
+
+echo "WORLD_SIZE: $WORLD_SIZE"
+echo "NPROC_PER_NODE: $NPROC_PER_NODE"
+
+# Training Arguments
+GLOBAL_BATCH_SIZE=128
+LOCAL_BATCH_SIZE=4
+GRADIENT_ACCUMULATION_STEPS=$[$GLOBAL_BATCH_SIZE/($WORLD_SIZE*$NPROC_PER_NODE*$LOCAL_BATCH_SIZE)]
+
+# Log Arguments
+export TRANSFORMERS_OFFLINE=1
+export WANDB_PROJECT=videollama2
+RUN_NAME=downstream_sft_settings
+DATA_DIR=datasets
+OUTP_DIR=work_dirs
+
+torchrun --nnodes $WORLD_SIZE \
+    --nproc_per_node $NPROC_PER_NODE \
+    --master_addr=$MASTER_ADDR \
+    --master_port=$MASTER_PORT \
+    --node_rank $RANK \
+    videollama2/train_flash_attn.py \
+    --deepspeed scripts/zero3.json \
+    --model_type videollama2 \
+    --model_path mistralai/Mistral-7B-Instruct-v0.2 \
+    --vision_tower openai/clip-vit-large-patch14-336 \
+    --mm_projector_type stc_connector \
+    --pretrain_mm_mlp_adapter DAMO-NLP-SG/VideoLLaMA2-7B-Base/mm_projector.bin \
+    --data_path   ${DATA_DIR}/videollava_sft/videochatgpt_llavaimage_tune.json \
+    --data_folder ${DATA_DIR}/videollava_sft/ \
+    --mm_vision_select_layer -2 \
+    --image_aspect_ratio pad \
+    --num_frames 8 \
+    --bf16 True \
+    --tf32 True \
+    --fp16 False \
+    --output_dir ${OUTP_DIR}/${WANDB_PROJECT}/finetune_${RUN_NAME} \
+    --num_train_epochs 1 \
+    --per_device_train_batch_size $LOCAL_BATCH_SIZE \
+    --per_device_eval_batch_size 4 \
+    --gradient_accumulation_steps $GRADIENT_ACCUMULATION_STEPS \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 500 \
+    --save_total_limit 99 \
+    --learning_rate 2e-5 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --model_max_length 2048 \
+    --gradient_checkpointing True \
+    --dataloader_num_workers 4 \
+    --report_to tensorboard \
+    --run_name $RUN_NAME \

scripts/custom/finetune_lora.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+# Environment Variables
+ARG_WORLD_SIZE=${1:-1}
+ARG_NPROC_PER_NODE=${2:-8}
+ARG_MASTER_ADDR="127.0.0.1"
+ARG_MASTER_PORT=16666
+ARG_RANK=0
+
+# Multiple conditions
+if [ ! -n "$WORLD_SIZE" ] || [ ! -n "$NPROC_PER_NODE" ]; then
+    WORLD_SIZE=$ARG_WORLD_SIZE
+    NPROC_PER_NODE=$ARG_NPROC_PER_NODE
+fi
+if [ ! -n "$MASTER_ADDR" ] || [ ! -n "$MASTER_PORT" ] || [ ! -n "$RANK" ]; then
+    MASTER_ADDR=$ARG_MASTER_ADDR
+    MASTER_PORT=$ARG_MASTER_PORT
+    RANK=$ARG_RANK
+fi
+
+echo "WORLD_SIZE: $WORLD_SIZE"
+echo "NPROC_PER_NODE: $NPROC_PER_NODE"
+
+# Training Arguments
+GLOBAL_BATCH_SIZE=128
+LOCAL_BATCH_SIZE=4
+GRADIENT_ACCUMULATION_STEPS=$[$GLOBAL_BATCH_SIZE/($WORLD_SIZE*$NPROC_PER_NODE*$LOCAL_BATCH_SIZE)]
+
+# Log Arguments
+export TRANSFORMERS_OFFLINE=1
+export WANDB_PROJECT=videollama2
+RUN_NAME=downstream_sft_settings_lora
+DATA_DIR=datasets
+OUTP_DIR=work_dirs
+
+torchrun --nnodes $WORLD_SIZE \
+    --nproc_per_node $NPROC_PER_NODE \
+    --master_addr=$MASTER_ADDR \
+    --master_port=$MASTER_PORT \
+    --node_rank $RANK \
+    videollama2/train_flash_attn.py \
+    --lora_enable True --lora_r 128 --lora_alpha 256 --mm_projector_lr 2e-5 \
+    --deepspeed scripts/zero3.json \
+    --model_type videollama2 \
+    --model_path mistralai/Mistral-7B-Instruct-v0.2 \
+    --vision_tower openai/clip-vit-large-patch14-336 \
+    --mm_projector_type stc_connector \
+    --pretrain_mm_mlp_adapter DAMO-NLP-SG/VideoLLaMA2-7B-Base/mm_projector.bin \
+    --data_path   ${DATA_DIR}/videollava_sft/videochatgpt_llavaimage_tune.json \
+    --data_folder ${DATA_DIR}/videollava_sft/ \
+    --mm_vision_select_layer -2 \
+    --image_aspect_ratio pad \
+    --num_frames 8 \
+    --bf16 True \
+    --tf32 True \
+    --fp16 False \
+    --output_dir ${OUTP_DIR}/${WANDB_PROJECT}/finetune_${RUN_NAME} \
+    --num_train_epochs 1 \
+    --per_device_train_batch_size $LOCAL_BATCH_SIZE \
+    --per_device_eval_batch_size 4 \
+    --gradient_accumulation_steps $GRADIENT_ACCUMULATION_STEPS \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 500 \
+    --save_total_limit 99 \
+    --learning_rate 2e-5 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --model_max_length 2048 \
+    --gradient_checkpointing True \
+    --dataloader_num_workers 4 \
+    --report_to tensorboard \
+    --run_name $RUN_NAME \

scripts/custom/finetune_qlora.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+# Environment Variables
+ARG_WORLD_SIZE=${1:-1}
+ARG_NPROC_PER_NODE=${2:-8}
+ARG_MASTER_ADDR="127.0.0.1"
+ARG_MASTER_PORT=16666
+ARG_RANK=0
+
+# Multiple conditions
+if [ ! -n "$WORLD_SIZE" ] || [ ! -n "$NPROC_PER_NODE" ]; then
+    WORLD_SIZE=$ARG_WORLD_SIZE
+    NPROC_PER_NODE=$ARG_NPROC_PER_NODE
+fi
+if [ ! -n "$MASTER_ADDR" ] || [ ! -n "$MASTER_PORT" ] || [ ! -n "$RANK" ]; then
+    MASTER_ADDR=$ARG_MASTER_ADDR
+    MASTER_PORT=$ARG_MASTER_PORT
+    RANK=$ARG_RANK
+fi
+
+echo "WORLD_SIZE: $WORLD_SIZE"
+echo "NPROC_PER_NODE: $NPROC_PER_NODE"
+
+# Training Arguments
+GLOBAL_BATCH_SIZE=128
+LOCAL_BATCH_SIZE=4
+GRADIENT_ACCUMULATION_STEPS=$[$GLOBAL_BATCH_SIZE/($WORLD_SIZE*$NPROC_PER_NODE*$LOCAL_BATCH_SIZE)]
+
+# Log Arguments
+export TRANSFORMERS_OFFLINE=1
+export WANDB_PROJECT=videollama2
+RUN_NAME=downstream_sft_settings_qlora
+DATA_DIR=datasets
+OUTP_DIR=work_dirs
+
+torchrun --nnodes $WORLD_SIZE \
+    --nproc_per_node $NPROC_PER_NODE \
+    --master_addr=$MASTER_ADDR \
+    --master_port=$MASTER_PORT \
+    --node_rank $RANK \
+    videollama2/train_flash_attn.py \
+    --lora_enable True --lora_r 128 --lora_alpha 256 --mm_projector_lr 2e-5 --bits 4 \
+    --deepspeed scripts/zero2.json \
+    --model_type videollama2 \
+    --model_path mistralai/Mistral-7B-Instruct-v0.2 \
+    --vision_tower openai/clip-vit-large-patch14-336 \
+    --mm_projector_type stc_connector \
+    --pretrain_mm_mlp_adapter DAMO-NLP-SG/VideoLLaMA2-7B-Base/mm_projector.bin \
+    --data_path   ${DATA_DIR}/videollava_sft/videochatgpt_llavaimage_tune.json \
+    --data_folder ${DATA_DIR}/videollava_sft/ \
+    --mm_vision_select_layer -2 \
+    --image_aspect_ratio pad \
+    --num_frames 8 \
+    --bf16 True \
+    --tf32 True \
+    --fp16 False \
+    --output_dir ${OUTP_DIR}/${WANDB_PROJECT}/finetune_${RUN_NAME} \
+    --num_train_epochs 1 \
+    --per_device_train_batch_size $LOCAL_BATCH_SIZE \
+    --per_device_eval_batch_size 4 \
+    --gradient_accumulation_steps $GRADIENT_ACCUMULATION_STEPS \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 500 \
+    --save_total_limit 99 \
+    --learning_rate 2e-5 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --model_max_length 2048 \
+    --gradient_checkpointing True \
+    --dataloader_num_workers 4 \
+    --report_to tensorboard \
+    --run_name $RUN_NAME \

scripts/eval/eval_video_cap_msvc.sh

@@ -0,0 +1,67 @@
+set -x
+
+EVAL_DATA_DIR=eval
+OUTPUT_DIR=eval_output
+CKPT=DAMO-NLP-SG/VideoLLaMA2-7B
+CKPT_NAME=$(echo $CKPT | rev | cut -d'/' -f1 | rev)
+
+gpu_list="${CUDA_VISIBLE_DEVICES:-0}"
+IFS=',' read -ra GPULIST <<< "$gpu_list"
+
+# divide data via the number of GPUs per task
+GPUS_PER_TASK=1
+CHUNKS=$((${#GPULIST[@]}/$GPUS_PER_TASK))
+
+output_file=${OUTPUT_DIR}/msvc/answers/${CKPT_NAME}/merge.json
+
+# judge if the number of json lines is 0
+if [ ! -f "$output_file" ] || [ $(cat "$output_file" | wc -l) -eq 0 ]; then
+    rm -f ${OUTPUT_DIR}/msvc/answers/${CKPT_NAME}/*.json
+fi
+
+if [ ! -f "$output_file" ]; then
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        # select the GPUs for the task
+        gpu_devices=$(IFS=,; echo "${GPULIST[*]:$(($IDX*$GPUS_PER_TASK)):$GPUS_PER_TASK}")
+        TRANSFORMERS_OFFLINE=1 CUDA_VISIBLE_DEVICES=${gpu_devices} python3 videollama2/eval/inference_video_cap_msvc.py \
+          --model-path ${CKPT} \
+          --video-folder ${EVAL_DATA_DIR}/msvc \
+          --question-file ${EVAL_DATA_DIR}/msvc/msvc.json \
+          --output-file ${OUTPUT_DIR}/msvc/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.json \
+          --num-chunks $CHUNKS \
+          --chunk-idx $IDX &
+    done
+
+    wait
+
+    # Clear out the output file if it exists.
+    > "$output_file"
+
+    #Loop through the indices and concatenate each file.
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        cat ${OUTPUT_DIR}/msvc/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.json >> "$output_file"
+    done
+fi
+
+
+AZURE_API_KEY=your_key
+AZURE_API_ENDPOINT=your_endpoint
+AZURE_API_DEPLOYNAME=your_deployname
+
+python3 videollama2/eval/eval_video_cap_msvc_correctness.py \
+    --pred-path $output_file \
+    --output-dir ${OUTPUT_DIR}/msvc/answers/${CKPT_NAME}/correctness_gpt \
+    --output-json ${OUTPUT_DIR}/msvc/answers/${CKPT_NAME}/correctness_results.json \
+    --api-key $AZURE_API_KEY \
+    --api-endpoint $AZURE_API_ENDPOINT \
+    --api-deployname $AZURE_API_DEPLOYNAME \
+    --num-tasks 4 \
+
+python3 videollama2/eval/eval_video_cap_msvc_detailedness.py \
+    --pred-path $output_file \
+    --output-dir ${OUTPUT_DIR}/msvc/answers/${CKPT_NAME}/detailedness_gpt \
+    --output-json ${OUTPUT_DIR}/msvc/answers/${CKPT_NAME}/detailedness_results.json \
+    --api-key $AZURE_API_KEY \
+    --api-endpoint $AZURE_API_ENDPOINT \
+    --api-deployname $AZURE_API_DEPLOYNAME \
+    --num-tasks 4 \

scripts/eval/eval_video_mcqa_egoschema.sh

@@ -0,0 +1,41 @@
+set -x
+
+EVAL_DATA_DIR=eval
+OUTPUT_DIR=eval_output
+CKPT=DAMO-NLP-SG/VideoLLaMA2-7B
+CKPT_NAME=$(echo $CKPT | rev | cut -d'/' -f1 | rev)
+
+gpu_list="${CUDA_VISIBLE_DEVICES:-0}"
+IFS=',' read -ra GPULIST <<< "$gpu_list"
+
+# divide data via the number of GPUs per task
+GPUS_PER_TASK=1
+CHUNKS=$((${#GPULIST[@]}/$GPUS_PER_TASK))
+
+output_file=${OUTPUT_DIR}/egoschema/answers/${CKPT_NAME}/merge.csv
+
+if [ ! -f "$output_file" ]; then
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        # select the GPUs for the task
+        gpu_devices=$(IFS=,; echo "${GPULIST[*]:$(($IDX*$GPUS_PER_TASK)):$GPUS_PER_TASK}")
+        TRANSFORMERS_OFFLINE=1 CUDA_VISIBLE_DEVICES=${gpu_devices} python3 videollama2/eval/inference_video_mcqa_egoschema.py \
+            --model-path ${CKPT} \
+            --video-folder ${EVAL_DATA_DIR}/egoschema/good_clips_git \
+            --question-file ${EVAL_DATA_DIR}/egoschema/questions.json \
+            --answer-file ${OUTPUT_DIR}/egoschema/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.csv \
+            --num-chunks $CHUNKS \
+            --chunk-idx $IDX &
+    done
+
+    wait
+
+    # Clear out the output file if it exists.
+    > "$output_file"
+
+    echo 'q_uid, answer' >> "$output_file"
+
+    # Loop through the indices and concatenate each file.
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        cat ${OUTPUT_DIR}/egoschema/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.csv >> "$output_file"
+    done
+fi

scripts/eval/eval_video_mcqa_mvbench.sh

@@ -0,0 +1,46 @@
+set -x
+
+EVAL_DATA_DIR=eval
+OUTPUT_DIR=eval_output
+CKPT=DAMO-NLP-SG/VideoLLaMA2-7B
+CKPT_NAME=$(echo $CKPT | rev | cut -d'/' -f1 | rev)
+
+gpu_list="${CUDA_VISIBLE_DEVICES:-0}"
+IFS=',' read -ra GPULIST <<< "$gpu_list"
+
+# divide data via the number of GPUs per task
+GPUS_PER_TASK=1
+CHUNKS=$((${#GPULIST[@]}/$GPUS_PER_TASK))
+
+output_file=${OUTPUT_DIR}/mvbench/answers/${CKPT_NAME}/merge.json
+
+# judge if the number of json lines is 0
+if [ ! -f "$output_file" ] || [ $(cat "$output_file" | wc -l) -eq 0 ]; then
+    rm -f ${OUTPUT_DIR}/mvbench/answers/${CKPT_NAME}/*.json
+fi
+
+if [ ! -f "$output_file" ]; then
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        gpu_devices=$(IFS=,; echo "${GPULIST[*]:$(($IDX*$GPUS_PER_TASK)):$GPUS_PER_TASK}")
+        TRANSFORMERS_OFFLINE=1 CUDA_VISIBLE_DEVICES=${gpu_devices} python3 videollama2/eval/inference_video_mcqa_mvbench.py \
+            --model-path ${CKPT} \
+            --video-folder ${EVAL_DATA_DIR}/mvbench/video \
+            --question-file ${EVAL_DATA_DIR}/mvbench/json \
+            --answer-file ${OUTPUT_DIR}/mvbench/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.json \
+            --num-chunks $CHUNKS \
+            --chunk-idx $IDX &
+    done
+
+    wait
+
+    # Clear out the output file if it exists.
+    > "$output_file"
+
+    # Loop through the indices and concatenate each file.
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        cat ${OUTPUT_DIR}/mvbench/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.json >> "$output_file"
+    done
+fi
+
+python3 videollama2/eval/eval_video_mcqa_mvbench.py \
+    --pred_path ${output_file} \

scripts/eval/eval_video_mcqa_perception_test_mcqa.sh

@@ -0,0 +1,45 @@
+set -x
+
+EVAL_DATA_DIR=eval
+OUTPUT_DIR=eval_output
+CKPT=DAMO-NLP-SG/VideoLLaMA2-7B
+CKPT_NAME=$(echo $CKPT | rev | cut -d'/' -f1 | rev)
+
+gpu_list="${CUDA_VISIBLE_DEVICES:-0}"
+IFS=',' read -ra GPULIST <<< "$gpu_list"
+
+# divide data via the number of GPUs per task
+GPUS_PER_TASK=1
+CHUNKS=$((${#GPULIST[@]}/$GPUS_PER_TASK))
+
+output_file=${OUTPUT_DIR}/perception_test_mcqa/answers/${CKPT_NAME}/merge.json
+
+if [ ! -f "$output_file" ]; then
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        # select the GPUs for the task
+        gpu_devices=$(IFS=,; echo "${GPULIST[*]:$(($IDX*$GPUS_PER_TASK)):$GPUS_PER_TASK}")    
+        TRANSFORMERS_OFFLINE=1 CUDA_VISIBLE_DEVICES=${gpu_devices} python3 videollama2/eval/inference_video_mcqa_perception_test_mcqa.py \
+            --model-path ${CKPT} \
+            --video-folder ${EVAL_DATA_DIR}/perception_test_mcqa/videos \
+            --question-file ${EVAL_DATA_DIR}/perception_test_mcqa/mc_question_test.json \
+            --answer-file ${OUTPUT_DIR}/perception_test_mcqa/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.json \
+            --num-chunks $CHUNKS \
+            --chunk-idx $IDX &
+    done
+
+    wait
+
+    # Clear out the output file if it exists.
+    > "$output_file"
+
+    echo "{" >> "$output_file"
+
+    # Loop through the indices and concatenate each file.
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        cat ${OUTPUT_DIR}/perception_test_mcqa/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.json >> "$output_file"
+    done
+
+    sed -i '$s/.$//' $output_file
+
+    echo "}" >> "$output_file"
+fi

scripts/eval/eval_video_mcqa_videomme.sh

@@ -0,0 +1,84 @@
+set -x
+
+EVAL_DATA_DIR=eval
+OUTPUT_DIR=eval_output
+CKPT=DAMO-NLP-SG/VideoLLaMA2-7B-16F
+CKPT_NAME=$(echo $CKPT | rev | cut -d'/' -f1 | rev)
+
+gpu_list="${CUDA_VISIBLE_DEVICES:-0}"
+IFS=',' read -ra GPULIST <<< "$gpu_list"
+
+# divide data via the number of GPUs per task
+GPUS_PER_TASK=1
+CHUNKS=$((${#GPULIST[@]}/$GPUS_PER_TASK))
+
+output_file=${OUTPUT_DIR}/videomme/answers/${CKPT_NAME}/merge.json
+output_sub_file=${OUTPUT_DIR}/videomme/answers/${CKPT_NAME}/merge_sub.json
+
+# judge if the number of json lines is 0
+if [ ! -f "$output_file" ] || [ $(cat "$output_file" | wc -l) -eq 0 ]; then
+    rm -f ${OUTPUT_DIR}/videomme/answers/${CKPT_NAME}/*.json
+fi
+
+
+if [ ! -f "$output_file" ]; then
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        # select the GPUs for the task
+        gpu_devices=$(IFS=,; echo "${GPULIST[*]:$(($IDX*$GPUS_PER_TASK)):$GPUS_PER_TASK}")
+        TRANSFORMERS_OFFLINE=1 CUDA_VISIBLE_DEVICES=${gpu_devices} python3 videollama2/eval/inference_video_mcqa_videomme.py \
+            --model-path ${CKPT} \
+            --video-folder ${EVAL_DATA_DIR}/videomme/videos \
+            --subtitle-folder ${EVAL_DATA_DIR}/videomme/subtitles \
+            --question-file ${EVAL_DATA_DIR}/videomme/test-00000-of-00001.parquet \
+            --answer-file ${OUTPUT_DIR}/videomme/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.json \
+            --num-chunks $CHUNKS \
+            --chunk-idx $IDX &
+    done
+
+    wait
+
+    # Clear out the output file if it exists.
+    > "$output_file"
+
+    echo "[" >> "$output_file"
+
+    #Loop through the indices and concatenate each file.
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        cat ${OUTPUT_DIR}/videomme/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.json >> "$output_file"
+    done
+
+    sed -i '$s/.$//' $output_file
+
+    echo "]" >> "$output_file"
+
+    # Clear out the output file if it exists.
+    > "$output_sub_file"
+
+    echo "[" >> "$output_sub_file"
+
+    #Loop through the indices and concatenate each file.
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        cat ${OUTPUT_DIR}/videomme/answers/${CKPT_NAME}/${CHUNKS}_${IDX}_sub.json >> "$output_sub_file"
+    done
+
+    sed -i '$s/.$//' $output_sub_file
+
+    echo "]" >> "$output_sub_file"
+fi
+
+
+python videollama2/eval/eval_video_mcqa_videomme.py \
+    --results_file $output_file \
+    --video_duration_type "short,medium,long" \
+    --return_categories_accuracy \
+    --return_sub_categories_accuracy \
+    --return_task_types_accuracy \
+    --skip_missing \
+
+python videollama2/eval/eval_video_mcqa_videomme.py \
+    --results_file $output_sub_file \
+    --video_duration_type "short,medium,long" \
+    --return_categories_accuracy \
+    --return_sub_categories_accuracy \
+    --return_task_types_accuracy \
+    --skip_missing \

scripts/eval/eval_video_oqa_activitynet.sh

@@ -0,0 +1,54 @@
+set -x
+
+EVAL_DATA_DIR=eval
+OUTPUT_DIR=eval_output
+# CKPT=DAMO-NLP-SG/VideoLLaMA2-7B
+CKPT=work_dirs/videollama2gemma2/finetune_2b_vllama2
+CKPT_NAME=$(echo $CKPT | rev | cut -d'/' -f1 | rev)
+
+gpu_list="${CUDA_VISIBLE_DEVICES:-0}"
+IFS=',' read -ra GPULIST <<< "$gpu_list"
+
+# divide data via the number of GPUs per task
+GPUS_PER_TASK=1
+CHUNKS=$((${#GPULIST[@]}/$GPUS_PER_TASK))
+
+output_file=${OUTPUT_DIR}/Activitynet_Zero_Shot_QA/answers/${CKPT_NAME}/merge.json
+
+if [ ! -f "$output_file" ]; then
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        # select the GPUs for the task
+        gpu_devices=$(IFS=,; echo "${GPULIST[*]:$(($IDX*$GPUS_PER_TASK)):$GPUS_PER_TASK}")
+        TRANSFORMERS_OFFLINE=1 CUDA_VISIBLE_DEVICES=${gpu_devices} python3 videollama2/eval/inference_video_oqa_activitynet.py \
+            --model-path ${CKPT} \
+            --video-folder ${EVAL_DATA_DIR}/Activitynet_Zero_Shot_QA/all_test \
+            --question-file ${EVAL_DATA_DIR}/Activitynet_Zero_Shot_QA/test_q.json \
+            --answer-file ${EVAL_DATA_DIR}/Activitynet_Zero_Shot_QA/test_a.json \
+            --output-file ${OUTPUT_DIR}/Activitynet_Zero_Shot_QA/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.json \
+            --num-chunks $CHUNKS \
+            --chunk-idx $IDX &
+    done
+
+    wait
+
+    # Clear out the output file if it exists.
+    > "$output_file"
+
+    #Loop through the indices and concatenate each file.
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        cat ${OUTPUT_DIR}/Activitynet_Zero_Shot_QA/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.json >> "$output_file"
+    done
+fi
+
+AZURE_API_KEY=your_key
+AZURE_API_ENDPOINT=your_endpoint
+AZURE_API_DEPLOYNAME=your_deployname
+
+python3 videollama2/eval/eval_video_oqa_activitynet.py \
+    --pred-path ${output_file} \
+    --output-dir ${OUTPUT_DIR}/Activitynet_Zero_Shot_QA/answers/${CKPT_NAME}/gpt \
+    --output-json ${OUTPUT_DIR}/Activitynet_Zero_Shot_QA/answers/${CKPT_NAME}/results.json \
+    --api-key $AZURE_API_KEY \
+    --api-endpoint $AZURE_API_ENDPOINT \
+    --api-deployname $AZURE_API_DEPLOYNAME \
+    --num-tasks 4

scripts/eval/eval_video_oqa_msvd.sh

@@ -0,0 +1,54 @@
+set -x
+
+EVAL_DATA_DIR=eval
+OUTPUT_DIR=eval_output
+CKPT=DAMO-NLP-SG/VideoLLaMA2-7B
+CKPT_NAME=$(echo $CKPT | rev | cut -d'/' -f1 | rev)
+
+gpu_list="${CUDA_VISIBLE_DEVICES:-0}"
+IFS=',' read -ra GPULIST <<< "$gpu_list"
+
+# divide data via the number of GPUs per task
+GPUS_PER_TASK=1
+CHUNKS=$((${#GPULIST[@]}/$GPUS_PER_TASK))
+
+output_file=${OUTPUT_DIR}/MSVD_Zero_Shot_QA/answers/${CKPT_NAME}/merge.json
+
+if [ ! -f "$output_file" ]; then
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        # select the GPUs for the task
+        gpu_devices=$(IFS=,; echo "${GPULIST[*]:$(($IDX*$GPUS_PER_TASK)):$GPUS_PER_TASK}")
+        TRANSFORMERS_OFFLINE=1 CUDA_VISIBLE_DEVICES=${gpu_devices} python3 videollama2/eval/inference_video_oqa_activitynet.py \
+            --model-path ${CKPT} \
+            --video-folder ${EVAL_DATA_DIR}/MSVD_Zero_Shot_QA/videos \
+            --question-file ${EVAL_DATA_DIR}/MSVD_Zero_Shot_QA/test_q.json \
+            --answer-file ${EVAL_DATA_DIR}/MSVD_Zero_Shot_QA/test_a.json \
+            --output-file ${OUTPUT_DIR}/MSVD_Zero_Shot_QA/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.json \
+            --num-chunks $CHUNKS \
+            --chunk-idx $IDX &
+    done
+
+    wait
+
+    # Clear out the output file if it exists.
+    > "$output_file"
+
+    #Loop through the indices and concatenate each file.
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        cat ${OUTPUT_DIR}/MSVD_Zero_Shot_QA/answers/${CKPT_NAME}/${CHUNKS}_${IDX}.json >> "$output_file"
+    done
+fi
+
+
+AZURE_API_KEY=your_key
+AZURE_API_ENDPOINT=your_endpoint
+AZURE_API_DEPLOYNAME=your_deployname
+
+python3 videollama2/eval/eval_video_oqa_activitynet.py \
+    --pred-path ${output_file} \
+    --output-dir ${OUTPUT_DIR}/MSVD_Zero_Shot_QA/answers/${CKPT_NAME}/gpt \
+    --output-json ${OUTPUT_DIR}/MSVD_Zero_Shot_QA/answers/${CKPT_NAME}/results.json \
+    --api-key $AZURE_API_KEY \
+    --api-endpoint $AZURE_API_ENDPOINT \
+    --api-deployname $AZURE_API_DEPLOYNAME \
+    --num-tasks 4

scripts/eval/eval_video_oqa_vcgpt_1_correctness.sh

@@ -0,0 +1,58 @@
+set -x
+
+EVAL_DATA_DIR=eval
+OUTPUT_DIR=eval_output
+CKPT=DAMO-NLP-SG/VideoLLaMA2-7B
+CKPT_NAME=$(echo $CKPT | rev | cut -d'/' -f1 | rev)
+
+gpu_list="${CUDA_VISIBLE_DEVICES:-0}"
+IFS=',' read -ra GPULIST <<< "$gpu_list"
+
+# divide data via the number of GPUs per task
+GPUS_PER_TASK=1
+CHUNKS=$((${#GPULIST[@]}/$GPUS_PER_TASK))
+
+output_file=${OUTPUT_DIR}/videochatgpt_gen/answers/correctness/${CKPT_NAME}/merge.json
+
+if [ ! -f "$output_file" ]; then
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        # select the GPUs for the task
+        gpu_devices=$(IFS=,; echo "${GPULIST[*]:$(($IDX*$GPUS_PER_TASK)):$GPUS_PER_TASK}")
+        TRANSFORMERS_OFFLINE=1 CUDA_VISIBLE_DEVICES=${gpu_devices} python3 videollama2/eval/inference_video_oqa_vcgpt_general.py \
+            --model-path ${CKPT} \
+            --video-folder ${EVAL_DATA_DIR}/videochatgpt_gen/Test_Videos \
+            --question-file ${EVAL_DATA_DIR}/videochatgpt_gen/generic_qa.json \
+            --answer-file ${OUTPUT_DIR}/videochatgpt_gen/answers/correctness/${CKPT_NAME}/${CHUNKS}_${IDX}.json \
+            --num-chunks $CHUNKS \
+            --chunk-idx $IDX &
+    done
+
+    wait
+
+    # Clear out the output file if it exists.
+    > "$output_file"
+
+    #Loop through the indices and concatenate each file.
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        cat ${OUTPUT_DIR}/videochatgpt_gen/answers/correctness/${CKPT_NAME}/${CHUNKS}_${IDX}.json >> "$output_file"
+    done
+
+    mkdir -p ${OUTPUT_DIR}/videochatgpt_gen/answers/detail/${CKPT_NAME}
+    mkdir -p ${OUTPUT_DIR}/videochatgpt_gen/answers/context/${CKPT_NAME}
+    cp ${output_file} ${OUTPUT_DIR}/videochatgpt_gen/answers/detail/${CKPT_NAME}/merge.json
+    cp ${output_file} ${OUTPUT_DIR}/videochatgpt_gen/answers/context/${CKPT_NAME}/merge.json
+fi
+
+
+AZURE_API_KEY=your_key
+AZURE_API_ENDPOINT=your_endpoint
+AZURE_API_DEPLOYNAME=your_deployname
+
+python3 videollama2/eval/eval_video_oqa_vcgpt_1_correctness.py \
+    --pred-path ${output_file} \
+    --output-dir ${OUTPUT_DIR}/videochatgpt_gen/answers/correctness/${CKPT_NAME}/gpt \
+    --output-json ${OUTPUT_DIR}/videochatgpt_gen/answers/correctness/${CKPT_NAME}/results.json \
+    --api-key $AZURE_API_KEY \
+    --api-endpoint $AZURE_API_ENDPOINT \
+    --api-deployname $AZURE_API_DEPLOYNAME \
+    --num-tasks 4

scripts/eval/eval_video_oqa_vcgpt_2_detail.sh

@@ -0,0 +1,58 @@
+set -x
+
+EVAL_DATA_DIR=eval
+OUTPUT_DIR=eval_output
+CKPT=DAMO-NLP-SG/VideoLLaMA2-7B
+CKPT_NAME=$(echo $CKPT | rev | cut -d'/' -f1 | rev)
+
+gpu_list="${CUDA_VISIBLE_DEVICES:-0}"
+IFS=',' read -ra GPULIST <<< "$gpu_list"
+
+# divide data via the number of GPUs per task
+GPUS_PER_TASK=1
+CHUNKS=$((${#GPULIST[@]}/$GPUS_PER_TASK))
+
+output_file=${OUTPUT_DIR}/videochatgpt_gen/answers/detail/${CKPT_NAME}/merge.json
+
+if [ ! -f "$output_file" ]; then
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        # select the GPUs for the task
+        gpu_devices=$(IFS=,; echo "${GPULIST[*]:$(($IDX*$GPUS_PER_TASK)):$GPUS_PER_TASK}")
+        TRANSFORMERS_OFFLINE=1 CUDA_VISIBLE_DEVICES=${gpu_devices} python3 videollama2/eval/run_inference_video_qa_gpt_general.py \
+            --model-path ${CKPT} \
+            --video-folder ${EVAL_DATA_DIR}/videochatgpt_gen/Test_Videos \
+            --question-file ${EVAL_DATA_DIR}/videochatgpt_gen/generic_qa.json \
+            --answer-file ${OUTPUT_DIR}/videochatgpt_gen/answers/detail/${CKPT_NAME}/${CHUNKS}_${IDX}.json \
+            --num-chunks $CHUNKS \
+            --chunk-idx $IDX &
+    done
+
+    wait
+
+    # Clear out the output file if it exists.
+    > "$output_file"
+
+    #Loop through the indices and concatenate each file.
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        cat ${OUTPUT_DIR}/videochatgpt_gen/answers/detail/${CKPT_NAME}/${CHUNKS}_${IDX}.json >> "$output_file"
+    done
+
+    mkdir -p ${OUTPUT_DIR}/videochatgpt_gen/answers/correctness/${CKPT_NAME}
+    mkdir -p ${OUTPUT_DIR}/videochatgpt_gen/answers/context/${CKPT_NAME}
+    cp ${output_file} ${OUTPUT_DIR}/videochatgpt_gen/answers/correctness/${CKPT_NAME}/merge.json
+    cp ${output_file} ${OUTPUT_DIR}/videochatgpt_gen/answers/context/${CKPT_NAME}/merge.json
+fi
+
+
+AZURE_API_KEY=your_key
+AZURE_API_ENDPOINT=your_endpoint
+AZURE_API_DEPLOYNAME=your_deployname
+
+python3 videollama2/eval/eval_video_oqa_vcgpt_2_detailed_orientation.py \
+    --pred-path ${output_file} \
+    --output-dir ${OUTPUT_DIR}/videochatgpt_gen/answers/detail/${CKPT_NAME}/gpt \
+    --output-json ${OUTPUT_DIR}/videochatgpt_gen/answers/detail/${CKPT_NAME}/results.json \
+    --api-key $AZURE_API_KEY \
+    --api-endpoint $AZURE_API_ENDPOINT \
+    --api-deployname $AZURE_API_DEPLOYNAME \
+    --num-tasks 4

scripts/eval/eval_video_oqa_vcgpt_3_context.sh

@@ -0,0 +1,58 @@
+set -x
+
+EVAL_DATA_DIR=eval
+OUTPUT_DIR=eval_output
+CKPT=DAMO-NLP-SG/VideoLLaMA2-7B
+CKPT_NAME=$(echo $CKPT | rev | cut -d'/' -f1 | rev)
+
+gpu_list="${CUDA_VISIBLE_DEVICES:-0}"
+IFS=',' read -ra GPULIST <<< "$gpu_list"
+
+# divide data via the number of GPUs per task
+GPUS_PER_TASK=1
+CHUNKS=$((${#GPULIST[@]}/$GPUS_PER_TASK))
+
+output_file=${OUTPUT_DIR}/videochatgpt_gen/answers/context/${CKPT_NAME}/merge.json
+
+if [ ! -f "$output_file" ]; then
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        # select the GPUs for the task
+        gpu_devices=$(IFS=,; echo "${GPULIST[*]:$(($IDX*$GPUS_PER_TASK)):$GPUS_PER_TASK}")
+        TRANSFORMERS_OFFLINE=1 CUDA_VISIBLE_DEVICES=${gpu_devices} python3 videollama2/eval/run_inference_video_qa_gpt_general.py \
+            --model-path ${CKPT} \
+            --video-folder ${EVAL_DATA_DIR}/videochatgpt_gen/Test_Videos \
+            --question-file ${EVAL_DATA_DIR}/videochatgpt_gen/generic_qa.json \
+            --answer-file ${OUTPUT_DIR}/videochatgpt_gen/answers/detail/${CKPT_NAME}/${CHUNKS}_${IDX}.json \
+            --num-chunks $CHUNKS \
+            --chunk-idx $IDX &
+    done
+
+    wait
+
+    # Clear out the output file if it exists.
+    > "$output_file"
+
+    #Loop through the indices and concatenate each file.
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        cat ${OUTPUT_DIR}/videochatgpt_gen/answers/context/${CKPT_NAME}/${CHUNKS}_${IDX}.json >> "$output_file"
+    done
+
+    mkdir -p ${OUTPUT_DIR}/videochatgpt_gen/answers/correctness/${CKPT_NAME}
+    mkdir -p ${OUTPUT_DIR}/videochatgpt_gen/answers/detail/${CKPT_NAME}
+    cp ${output_file} ${OUTPUT_DIR}/videochatgpt_gen/answers/correctness/${CKPT_NAME}/merge.json
+    cp ${output_file} ${OUTPUT_DIR}/videochatgpt_gen/answers/detail/${CKPT_NAME}/merge.json
+fi
+
+
+AZURE_API_KEY=your_key
+AZURE_API_ENDPOINT=your_endpoint
+AZURE_API_DEPLOYNAME=your_deployname
+
+python3 videollama2/eval/eval_video_oqa_vcgpt_3_context.py \
+    --pred-path ${output_file} \
+    --output-dir ${OUTPUT_DIR}/videochatgpt_gen/answers/context/${CKPT_NAME}/gpt \
+    --output-json ${OUTPUT_DIR}/videochatgpt_gen/answers/context/${CKPT_NAME}/results.json \
+    --api-key $AZURE_API_KEY \
+    --api-endpoint $AZURE_API_ENDPOINT \
+    --api-deployname $AZURE_API_DEPLOYNAME \
+    --num-tasks 4

scripts/eval/eval_video_oqa_vcgpt_4_temporal.sh

@@ -0,0 +1,54 @@
+set -x
+
+EVAL_DATA_DIR=eval
+OUTPUT_DIR=eval_output
+CKPT=DAMO-NLP-SG/VideoLLaMA2-7B
+CKPT_NAME=$(echo $CKPT | rev | cut -d'/' -f1 | rev)
+
+gpu_list="${CUDA_VISIBLE_DEVICES:-0}"
+IFS=',' read -ra GPULIST <<< "$gpu_list"
+
+# divide data via the number of GPUs per task
+GPUS_PER_TASK=1
+CHUNKS=$((${#GPULIST[@]}/$GPUS_PER_TASK))
+
+output_file=${OUTPUT_DIR}/videochatgpt_gen/answers/temporal/${CKPT_NAME}/merge.json
+
+# if output_file not exists then inference
+if [ ! -f "$output_file" ]; then
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        # select the GPUs for the task
+        gpu_devices=$(IFS=,; echo "${GPULIST[*]:$(($IDX*$GPUS_PER_TASK)):$GPUS_PER_TASK}")
+        TRANSFORMERS_OFFLINE=1 CUDA_VISIBLE_DEVICES=${gpu_devices} python3 videollama2/eval/inference_video_oqa_vcgpt_general.py \
+            --model-path ${CKPT} \
+            --video-folder ${EVAL_DATA_DIR}/videochatgpt_gen/Test_Videos \
+            --question-file ${EVAL_DATA_DIR}/videochatgpt_gen/temporal_qa.json \
+            --answer-file ${OUTPUT_DIR}/videochatgpt_gen/answers/temporal/${CKPT_NAME}/${CHUNKS}_${IDX}.json \
+            --num-chunks $CHUNKS \
+            --chunk-idx $IDX &
+    done
+
+    wait
+
+    # Clear out the output file if it exists.
+    > "$output_file"
+
+    #Loop through the indices and concatenate each file.
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        cat ${OUTPUT_DIR}/videochatgpt_gen/answers/temporal/${CKPT_NAME}/${CHUNKS}_${IDX}.json >> "$output_file"
+    done
+fi
+
+
+AZURE_API_KEY=your_key
+AZURE_API_ENDPOINT=your_endpoint
+AZURE_API_DEPLOYNAME=your_deployname
+
+python3 videollama2/eval/eval_video_oqa_vcgpt_4_temporal.py \
+    --pred-path ${output_file} \
+    --output-dir ${OUTPUT_DIR}/videochatgpt_gen/answers/temporal/${CKPT_NAME}/gpt \
+    --output-json ${OUTPUT_DIR}/videochatgpt_gen/answers/temporal/${CKPT_NAME}/results.json \
+    --api-key $AZURE_API_KEY \
+    --api-endpoint $AZURE_API_ENDPOINT \
+    --api-deployname $AZURE_API_DEPLOYNAME \
+    --num-tasks 4

scripts/eval/eval_video_oqa_vcgpt_5_consistency.sh

@@ -0,0 +1,54 @@
+set -x
+
+EVAL_DATA_DIR=eval
+OUTPUT_DIR=eval_output
+CKPT=DAMO-NLP-SG/VideoLLaMA2-7B
+CKPT_NAME=$(echo $CKPT | rev | cut -d'/' -f1 | rev)
+
+gpu_list="${CUDA_VISIBLE_DEVICES:-0}"
+IFS=',' read -ra GPULIST <<< "$gpu_list"
+
+# divide data via the number of GPUs per task
+GPUS_PER_TASK=1
+CHUNKS=$((${#GPULIST[@]}/$GPUS_PER_TASK))
+
+output_file=${OUTPUT_DIR}/videochatgpt_gen/answers/consistency/${CKPT_NAME}/merge.json
+
+# if output_file not exists then inference
+if [ ! -f "$output_file" ]; then
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        # select the GPUs for the task
+        gpu_devices=$(IFS=,; echo "${GPULIST[*]:$(($IDX*$GPUS_PER_TASK)):$GPUS_PER_TASK}")
+        TRANSFORMERS_OFFLINE=1 CUDA_VISIBLE_DEVICES=${gpu_devices} python3 videollama2/eval/inference_video_oqa_vcgpt_consistency.py \
+            --model-path ${CKPT} \
+            --video-folder ${EVAL_DATA_DIR}/videochatgpt_gen/Test_Videos \
+            --question-file ${EVAL_DATA_DIR}/videochatgpt_gen/consistency_qa.json \
+            --answer-file ${OUTPUT_DIR}/videochatgpt_gen/answers/consistency/${CKPT_NAME}/${CHUNKS}_${IDX}.json \
+            --num-chunks $CHUNKS \
+            --chunk-idx $IDX &
+    done
+
+    wait
+
+    # Clear out the output file if it exists.
+    > "$output_file"
+
+    #Loop through the indices and concatenate each file.
+    for IDX in $(seq 0 $((CHUNKS-1))); do
+        cat ${OUTPUT_DIR}/videochatgpt_gen/answers/consistency/${CKPT_NAME}/${CHUNKS}_${IDX}.json >> "$output_file"
+    done
+fi
+
+
+AZURE_API_KEY=your_key
+AZURE_API_ENDPOINT=your_endpoint
+AZURE_API_DEPLOYNAME=your_deployname
+
+python3 videollama2/eval/eval_video_oqa_vcgpt_5_consistency.py \
+    --pred-path ${output_file} \
+    --output-dir ${OUTPUT_DIR}/videochatgpt_gen/answers/consistency/${CKPT_NAME}/gpt \
+    --output-json ${OUTPUT_DIR}/videochatgpt_gen/answers/consistency/${CKPT_NAME}/results.json \
+    --api-key $AZURE_API_KEY \
+    --api-endpoint $AZURE_API_ENDPOINT \
+    --api-deployname $AZURE_API_DEPLOYNAME \
+    --num-tasks 4

scripts/siglip/finetune_gemma2.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+# Environment Variables
+ARG_WORLD_SIZE=${1:-1}
+ARG_NPROC_PER_NODE=${2:-8}
+ARG_MASTER_ADDR="127.0.0.1"
+ARG_MASTER_PORT=16667
+ARG_RANK=0
+
+# Multiple conditions
+if [ ! -n "$WORLD_SIZE" ] || [ ! -n "$NPROC_PER_NODE" ]; then
+    WORLD_SIZE=$ARG_WORLD_SIZE
+    NPROC_PER_NODE=$ARG_NPROC_PER_NODE
+fi
+if [ ! -n "$MASTER_ADDR" ] || [ ! -n "$MASTER_PORT" ] || [ ! -n "$RANK" ]; then
+    MASTER_ADDR=$ARG_MASTER_ADDR
+    MASTER_PORT=$ARG_MASTER_PORT
+    RANK=$ARG_RANK
+fi
+
+echo "WORLD_SIZE: $WORLD_SIZE"
+echo "NPROC_PER_NODE: $NPROC_PER_NODE"
+
+# Training Arguments
+GLOBAL_BATCH_SIZE=128
+LOCAL_BATCH_SIZE=4
+GRADIENT_ACCUMULATION_STEPS=$[$GLOBAL_BATCH_SIZE/($WORLD_SIZE*$NPROC_PER_NODE*$LOCAL_BATCH_SIZE)]
+echo $GRADIENT_ACCUMULATION_STEPS
+
+# Log Arguments
+export TRANSFORMERS_OFFLINE=1
+export WANDB_PROJECT=videollama2gemma2_siglip
+RUN_NAME=vllava_settings
+DATA_DIR=datasets
+OUTP_DIR=work_dirs
+
+torchrun --nnodes $WORLD_SIZE \
+    --nproc_per_node $NPROC_PER_NODE \
+    --master_addr=$MASTER_ADDR \
+    --master_port=$MASTER_PORT \
+    --node_rank $RANK \
+    videollama2/train_flash_attn.py \
+    --deepspeed scripts/zero3.json \
+    --model_type videollama2_gemma2 \
+    --model_path google/gemma-2-2b-it \
+    --vision_tower google/siglip-so400m-patch14-384 \
+    --mm_projector_type stc_connector_v35 \
+    --pretrain_mm_mlp_adapter ${OUTP_DIR}/${WANDB_PROJECT}/pretrain_${RUN_NAME}/mm_projector.bin \
+    --data_path   ${DATA_DIR}/videollava_sft/videochatgpt_llavaimage_tune.json \
+    --data_folder ${DATA_DIR}/videollava_sft/ \
+    --mm_vision_select_layer -2 \
+    --image_aspect_ratio pad \
+    --num_frames 8 \
+    --bf16 True \
+    --tf32 True \
+    --fp16 False \
+    --output_dir ${OUTP_DIR}/${WANDB_PROJECT}/finetune_${RUN_NAME} \
+    --num_train_epochs 3 \
+    --per_device_train_batch_size $LOCAL_BATCH_SIZE \
+    --per_device_eval_batch_size 4 \
+    --gradient_accumulation_steps $GRADIENT_ACCUMULATION_STEPS \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 200 \
+    --save_total_limit 99 \
+    --learning_rate 2e-5 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --model_max_length 2048 \
+    --gradient_checkpointing True \
+    --dataloader_num_workers 4 \
+    --report_to tensorboard \
+    --run_name finetune_$RUN_NAME \

scripts/siglip/finetune_mistral.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+# Environment Variables
+ARG_WORLD_SIZE=${1:-1}
+ARG_NPROC_PER_NODE=${2:-8}
+ARG_MASTER_ADDR="127.0.0.1"
+ARG_MASTER_PORT=16667
+ARG_RANK=0
+
+# Multiple conditions
+if [ ! -n "$WORLD_SIZE" ] || [ ! -n "$NPROC_PER_NODE" ]; then
+    WORLD_SIZE=$ARG_WORLD_SIZE
+    NPROC_PER_NODE=$ARG_NPROC_PER_NODE
+fi
+if [ ! -n "$MASTER_ADDR" ] || [ ! -n "$MASTER_PORT" ] || [ ! -n "$RANK" ]; then
+    MASTER_ADDR=$ARG_MASTER_ADDR
+    MASTER_PORT=$ARG_MASTER_PORT
+    RANK=$ARG_RANK
+fi
+
+echo "WORLD_SIZE: $WORLD_SIZE"
+echo "NPROC_PER_NODE: $NPROC_PER_NODE"
+
+# Training Arguments
+GLOBAL_BATCH_SIZE=128
+LOCAL_BATCH_SIZE=4
+GRADIENT_ACCUMULATION_STEPS=$[$GLOBAL_BATCH_SIZE/($WORLD_SIZE*$NPROC_PER_NODE*$LOCAL_BATCH_SIZE)]
+echo $GRADIENT_ACCUMULATION_STEPS
+
+# Log Arguments
+export TRANSFORMERS_OFFLINE=1
+export WANDB_PROJECT=videollama2mistral_siglip
+RUN_NAME=vllava_settings
+DATA_DIR=datasets
+OUTP_DIR=work_dirs
+
+torchrun --nnodes $WORLD_SIZE \
+    --nproc_per_node $NPROC_PER_NODE \
+    --master_addr=$MASTER_ADDR \
+    --master_port=$MASTER_PORT \
+    --node_rank $RANK \
+    videollama2/train_flash_attn.py \
+    --deepspeed scripts/zero3.json \
+    --model_type videollama2 \
+    --model_path mistralai/Mistral-7B-Instruct-v0.2 \
+    --vision_tower google/siglip-so400m-patch14-384 \
+    --mm_projector_type stc_connector_v35 \
+    --pretrain_mm_mlp_adapter ${OUTP_DIR}/${WANDB_PROJECT}/pretrain_${RUN_NAME}/mm_projector.bin \
+    --data_path   ${DATA_DIR}/videollava_sft/videochatgpt_llavaimage_tune.json \
+    --data_folder ${DATA_DIR}/videollava_sft/ \
+    --mm_vision_select_layer -2 \
+    --image_aspect_ratio pad \
+    --num_frames 8 \
+    --bf16 True \
+    --tf32 True \
+    --fp16 False \
+    --output_dir ${OUTP_DIR}/${WANDB_PROJECT}/finetune_${RUN_NAME} \
+    --num_train_epochs 3 \
+    --per_device_train_batch_size $LOCAL_BATCH_SIZE \
+    --per_device_eval_batch_size 4 \
+    --gradient_accumulation_steps $GRADIENT_ACCUMULATION_STEPS \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 200 \
+    --save_total_limit 99 \
+    --learning_rate 2e-5 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --model_max_length 2048 \
+    --gradient_checkpointing True \
+    --dataloader_num_workers 4 \
+    --report_to wandb \
+    --run_name finetune_$RUN_NAME \

scripts/siglip/finetune_phi3.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+# Environment Variables
+ARG_WORLD_SIZE=${1:-1}
+ARG_NPROC_PER_NODE=${2:-8}
+ARG_MASTER_ADDR="127.0.0.1"
+ARG_MASTER_PORT=16667
+ARG_RANK=0
+
+# Multiple conditions
+if [ ! -n "$WORLD_SIZE" ] || [ ! -n "$NPROC_PER_NODE" ]; then
+    WORLD_SIZE=$ARG_WORLD_SIZE
+    NPROC_PER_NODE=$ARG_NPROC_PER_NODE
+fi
+if [ ! -n "$MASTER_ADDR" ] || [ ! -n "$MASTER_PORT" ] || [ ! -n "$RANK" ]; then
+    MASTER_ADDR=$ARG_MASTER_ADDR
+    MASTER_PORT=$ARG_MASTER_PORT
+    RANK=$ARG_RANK
+fi
+
+echo "WORLD_SIZE: $WORLD_SIZE"
+echo "NPROC_PER_NODE: $NPROC_PER_NODE"
+
+# Training Arguments
+GLOBAL_BATCH_SIZE=128
+LOCAL_BATCH_SIZE=4
+GRADIENT_ACCUMULATION_STEPS=$[$GLOBAL_BATCH_SIZE/($WORLD_SIZE*$NPROC_PER_NODE*$LOCAL_BATCH_SIZE)]
+echo $GRADIENT_ACCUMULATION_STEPS
+
+# Log Arguments
+export TRANSFORMERS_OFFLINE=1
+export WANDB_PROJECT=videollama2phi3_siglip
+RUN_NAME=vllava_settings
+DATA_DIR=datasets
+OUTP_DIR=work_dirs
+
+torchrun --nnodes $WORLD_SIZE \
+    --nproc_per_node $NPROC_PER_NODE \
+    --master_addr=$MASTER_ADDR \
+    --master_port=$MASTER_PORT \
+    --node_rank $RANK \
+    videollama2/train_flash_attn.py \
+    --deepspeed scripts/zero3.json \
+    --model_type videollama2_phi3 \
+    --model_path microsoft/Phi-3-mini-4k-instruct \
+    --vision_tower google/siglip-so400m-patch14-384 \
+    --mm_projector_type stc_connector_v35 \
+    --pretrain_mm_mlp_adapter ${OUTP_DIR}/${WANDB_PROJECT}/pretrain_${RUN_NAME}/mm_projector.bin \
+    --data_path   ${DATA_DIR}/videollava_sft/videochatgpt_llavaimage_tune.json \
+    --data_folder ${DATA_DIR}/videollava_sft/ \
+    --mm_vision_select_layer -2 \
+    --image_aspect_ratio pad \
+    --num_frames 8 \
+    --bf16 True \
+    --tf32 True \
+    --fp16 False \
+    --output_dir ${OUTP_DIR}/${WANDB_PROJECT}/finetune_${RUN_NAME} \
+    --num_train_epochs 3 \
+    --per_device_train_batch_size $LOCAL_BATCH_SIZE \
+    --per_device_eval_batch_size 4 \
+    --gradient_accumulation_steps $GRADIENT_ACCUMULATION_STEPS \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 200 \
+    --save_total_limit 99 \
+    --learning_rate 2e-5 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --model_max_length 2048 \
+    --gradient_checkpointing True \
+    --dataloader_num_workers 4 \
+    --report_to tensorboard \
+    --run_name finetune_$RUN_NAME \

scripts/siglip/finetune_qwen2.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+# Environment Variables
+ARG_WORLD_SIZE=${1:-1}
+ARG_NPROC_PER_NODE=${2:-8}
+ARG_MASTER_ADDR="127.0.0.1"
+ARG_MASTER_PORT=16666
+ARG_RANK=0
+
+# Multiple conditions
+if [ ! -n "$WORLD_SIZE" ] || [ ! -n "$NPROC_PER_NODE" ]; then
+    WORLD_SIZE=$ARG_WORLD_SIZE
+    NPROC_PER_NODE=$ARG_NPROC_PER_NODE
+fi
+if [ ! -n "$MASTER_ADDR" ] || [ ! -n "$MASTER_PORT" ] || [ ! -n "$RANK" ]; then
+    MASTER_ADDR=$ARG_MASTER_ADDR
+    MASTER_PORT=$ARG_MASTER_PORT
+    RANK=$ARG_RANK
+fi
+
+echo "WORLD_SIZE: $WORLD_SIZE"
+echo "NPROC_PER_NODE: $NPROC_PER_NODE"
+
+# Training Arguments
+GLOBAL_BATCH_SIZE=128
+LOCAL_BATCH_SIZE=4
+GRADIENT_ACCUMULATION_STEPS=$[$GLOBAL_BATCH_SIZE/($WORLD_SIZE*$NPROC_PER_NODE*$LOCAL_BATCH_SIZE)]
+echo $GRADIENT_ACCUMULATION_STEPS
+
+# Log Arguments
+export TRANSFORMERS_OFFLINE=1
+export WANDB_PROJECT=videollama2qwen2_siglip
+RUN_NAME=vllava_settings
+DATA_DIR=datasets
+OUTP_DIR=work_dirs
+
+torchrun --nnodes $WORLD_SIZE \
+    --nproc_per_node $NPROC_PER_NODE \
+    --master_addr=$MASTER_ADDR \
+    --master_port=$MASTER_PORT \
+    --node_rank $RANK \
+    videollama2/train_flash_attn.py \
+    --deepspeed scripts/zero3.json \
+    --model_type videollama2_qwen2 \
+    --model_path Qwen/Qwen2-7B-Instruct \
+    --vision_tower google/siglip-so400m-patch14-384 \
+    --mm_projector_type stc_connector_v35 \
+    --pretrain_mm_mlp_adapter ${OUTP_DIR}/${WANDB_PROJECT}/pretrain_${RUN_NAME}/mm_projector.bin \
+    --data_path   ${DATA_DIR}/videollava_sft/videochatgpt_llavaimage_tune.json \
+    --data_folder ${DATA_DIR}/videollava_sft/ \
+    --mm_vision_select_layer -2 \
+    --image_aspect_ratio pad \
+    --num_frames 8 \
+    --bf16 True \
+    --tf32 True \
+    --fp16 False \
+    --output_dir ${OUTP_DIR}/${WANDB_PROJECT}/finetune_${RUN_NAME} \
+    --num_train_epochs 1 \
+    --per_device_train_batch_size $LOCAL_BATCH_SIZE \
+    --per_device_eval_batch_size 4 \
+    --gradient_accumulation_steps $GRADIENT_ACCUMULATION_STEPS \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 500 \
+    --save_total_limit 99 \
+    --learning_rate 2e-5 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --model_max_length 2048 \
+    --gradient_checkpointing True \
+    --dataloader_num_workers 4 \
+    --report_to tensorboard \
+    --run_name $RUN_NAME \

scripts/siglip/pretrain_gemma2.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+# Environment Variables
+ARG_WORLD_SIZE=${1:-1}
+ARG_NPROC_PER_NODE=${2:-8}
+ARG_MASTER_ADDR="127.0.0.1"
+ARG_MASTER_PORT=16666
+ARG_RANK=0
+
+# Multiple conditions
+if [ ! -n "$WORLD_SIZE" ] || [ ! -n "$NPROC_PER_NODE" ]; then
+    WORLD_SIZE=$ARG_WORLD_SIZE
+    NPROC_PER_NODE=$ARG_NPROC_PER_NODE
+fi
+if [ ! -n "$MASTER_ADDR" ] || [ ! -n "$MASTER_PORT" ] || [ ! -n "$RANK" ]; then
+    MASTER_ADDR=$ARG_MASTER_ADDR
+    MASTER_PORT=$ARG_MASTER_PORT
+    RANK=$ARG_RANK
+fi
+
+echo "WORLD_SIZE: $WORLD_SIZE"
+echo "NPROC_PER_NODE: $NPROC_PER_NODE"
+
+# Training Arguments
+GLOBAL_BATCH_SIZE=256
+LOCAL_BATCH_SIZE=4
+GRADIENT_ACCUMULATION_STEPS=$[$GLOBAL_BATCH_SIZE/($WORLD_SIZE*$NPROC_PER_NODE*$LOCAL_BATCH_SIZE)]
+echo $GRADIENT_ACCUMULATION_STEPS
+
+# Log Arguments
+export TRANSFORMERS_OFFLINE=1
+export WANDB_PROJECT=videollama2gemma2_siglip
+RUN_NAME=vllava_settings
+DATA_DIR=datasets
+OUTP_DIR=work_dirs
+
+torchrun --nnodes $WORLD_SIZE \
+    --nproc_per_node $NPROC_PER_NODE  \
+    --master_addr=$MASTER_ADDR \
+    --master_port=$MASTER_PORT \
+    --node_rank $RANK \
+    videollama2/train_flash_attn.py \
+    --deepspeed scripts/zero3.json \
+    --model_type videollama2_gemma2 \
+    --model_path google/gemma-2-2b-it \
+    --vision_tower google/siglip-so400m-patch14-384 \
+    --mm_projector_type stc_connector_v35 \
+    --tune_mm_mlp_adapter True \
+    --data_path   ${DATA_DIR}/videollava_pt/valley_llavaimage.json \
+    --data_folder ${DATA_DIR}/videollava_pt/ \
+    --mm_vision_select_layer -2 \
+    --num_frames 8 \
+    --bf16 True \
+    --tf32 True \
+    --fp16 False \
+    --output_dir ${OUTP_DIR}/${WANDB_PROJECT}/pretrain_${RUN_NAME} \
+    --num_train_epochs 1 \
+    --per_device_train_batch_size $LOCAL_BATCH_SIZE \
+    --per_device_eval_batch_size 4 \
+    --gradient_accumulation_steps $GRADIENT_ACCUMULATION_STEPS \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 500 \
+    --save_total_limit 99 \
+    --learning_rate 1e-3 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --model_max_length 2048 \
+    --gradient_checkpointing True \
+    --dataloader_num_workers 4 \
+    --lazy_preprocess True \
+    --report_to tensorboard \
+    --run_name pretrain_$RUN_NAME \

scripts/siglip/pretrain_mistral.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+# Environment Variables
+ARG_WORLD_SIZE=${1:-1}
+ARG_NPROC_PER_NODE=${2:-8}
+ARG_MASTER_ADDR="127.0.0.1"
+ARG_MASTER_PORT=16666
+ARG_RANK=0
+
+# Multiple conditions
+if [ ! -n "$WORLD_SIZE" ] || [ ! -n "$NPROC_PER_NODE" ]; then
+    WORLD_SIZE=$ARG_WORLD_SIZE
+    NPROC_PER_NODE=$ARG_NPROC_PER_NODE
+fi
+if [ ! -n "$MASTER_ADDR" ] || [ ! -n "$MASTER_PORT" ] || [ ! -n "$RANK" ]; then
+    MASTER_ADDR=$ARG_MASTER_ADDR
+    MASTER_PORT=$ARG_MASTER_PORT
+    RANK=$ARG_RANK
+fi
+
+echo "WORLD_SIZE: $WORLD_SIZE"
+echo "NPROC_PER_NODE: $NPROC_PER_NODE"
+
+# Training Arguments
+GLOBAL_BATCH_SIZE=256
+LOCAL_BATCH_SIZE=8
+GRADIENT_ACCUMULATION_STEPS=$[$GLOBAL_BATCH_SIZE/($WORLD_SIZE*$NPROC_PER_NODE*$LOCAL_BATCH_SIZE)]
+echo $GRADIENT_ACCUMULATION_STEPS
+
+# Log Arguments
+export TRANSFORMERS_OFFLINE=1
+export WANDB_PROJECT=videollama2mistral_siglip
+RUN_NAME=vllava_settings
+DATA_DIR=datasets
+OUTP_DIR=work_dirs
+
+torchrun --nnodes $WORLD_SIZE \
+    --nproc_per_node $NPROC_PER_NODE  \
+    --master_addr=$MASTER_ADDR \
+    --master_port=$MASTER_PORT \
+    --node_rank $RANK \
+    videollama2/train_flash_attn.py \
+    --deepspeed scripts/zero3.json \
+    --model_type videollama2 \
+    --model_path mistralai/Mistral-7B-Instruct-v0.2 \
+    --vision_tower google/siglip-so400m-patch14-384 \
+    --mm_projector_type stc_connector_v35 \
+    --tune_mm_mlp_adapter True \
+    --data_path   ${DATA_DIR}/videollava_pt/valley_llavaimage.json \
+    --data_folder ${DATA_DIR}/videollava_pt/ \
+    --mm_vision_select_layer -2 \
+    --num_frames 8 \
+    --bf16 True \
+    --tf32 True \
+    --fp16 False \
+    --output_dir ${OUTP_DIR}/${WANDB_PROJECT}/pretrain_${RUN_NAME} \
+    --num_train_epochs 1 \
+    --per_device_train_batch_size $LOCAL_BATCH_SIZE \
+    --per_device_eval_batch_size 4 \
+    --gradient_accumulation_steps $GRADIENT_ACCUMULATION_STEPS \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 500 \
+    --save_total_limit 99 \
+    --learning_rate 1e-3 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --model_max_length 2048 \
+    --gradient_checkpointing True \
+    --dataloader_num_workers 16 \
+    --lazy_preprocess True \
+    --report_to tensorboard \
+    --run_name pretrain_$RUN_NAME \

scripts/siglip/pretrain_phi3.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+# Environment Variables
+ARG_WORLD_SIZE=${1:-1}
+ARG_NPROC_PER_NODE=${2:-8}
+ARG_MASTER_ADDR="127.0.0.1"
+ARG_MASTER_PORT=16666
+ARG_RANK=0
+
+# Multiple conditions
+if [ ! -n "$WORLD_SIZE" ] || [ ! -n "$NPROC_PER_NODE" ]; then
+    WORLD_SIZE=$ARG_WORLD_SIZE
+    NPROC_PER_NODE=$ARG_NPROC_PER_NODE
+fi
+if [ ! -n "$MASTER_ADDR" ] || [ ! -n "$MASTER_PORT" ] || [ ! -n "$RANK" ]; then
+    MASTER_ADDR=$ARG_MASTER_ADDR
+    MASTER_PORT=$ARG_MASTER_PORT
+    RANK=$ARG_RANK
+fi
+
+echo "WORLD_SIZE: $WORLD_SIZE"
+echo "NPROC_PER_NODE: $NPROC_PER_NODE"
+
+# Training Arguments
+GLOBAL_BATCH_SIZE=256
+LOCAL_BATCH_SIZE=8
+GRADIENT_ACCUMULATION_STEPS=$[$GLOBAL_BATCH_SIZE/($WORLD_SIZE*$NPROC_PER_NODE*$LOCAL_BATCH_SIZE)]
+echo $GRADIENT_ACCUMULATION_STEPS
+
+# Log Arguments
+export TRANSFORMERS_OFFLINE=1
+export WANDB_PROJECT=videollama2phi3_siglip
+RUN_NAME=vllava_settings
+DATA_DIR=datasets
+OUTP_DIR=work_dirs
+
+torchrun --nnodes $WORLD_SIZE \
+    --nproc_per_node $NPROC_PER_NODE  \
+    --master_addr=$MASTER_ADDR \
+    --master_port=$MASTER_PORT \
+    --node_rank $RANK \
+    videollama2/train_flash_attn.py \
+    --deepspeed scripts/zero3.json \
+    --model_type videollama2_phi3 \
+    --model_path microsoft/Phi-3-mini-4k-instruct \
+    --vision_tower google/siglip-so400m-patch14-384 \
+    --mm_projector_type stc_connector_v35 \
+    --tune_mm_mlp_adapter True \
+    --data_path   ${DATA_DIR}/videollava_pt/valley_llavaimage.json \
+    --data_folder ${DATA_DIR}/videollava_pt/ \
+    --mm_vision_select_layer -2 \
+    --num_frames 8 \
+    --bf16 True \
+    --tf32 True \
+    --fp16 False \
+    --output_dir ${OUTP_DIR}/${WANDB_PROJECT}/pretrain_${RUN_NAME} \
+    --num_train_epochs 1 \
+    --per_device_train_batch_size $LOCAL_BATCH_SIZE \
+    --per_device_eval_batch_size 4 \
+    --gradient_accumulation_steps $GRADIENT_ACCUMULATION_STEPS \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 500 \
+    --save_total_limit 99 \
+    --learning_rate 1e-3 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --model_max_length 2048 \
+    --gradient_checkpointing True \
+    --dataloader_num_workers 4 \
+    --lazy_preprocess True \
+    --report_to tensorboard \
+    --run_name pretrain_$RUN_NAME \

scripts/siglip/pretrain_qwen2.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+# Environment Variables
+ARG_WORLD_SIZE=${1:-1}
+ARG_NPROC_PER_NODE=${2:-8}
+ARG_MASTER_ADDR="127.0.0.1"
+ARG_MASTER_PORT=16666
+ARG_RANK=0
+
+# Multiple conditions
+if [ ! -n "$WORLD_SIZE" ] || [ ! -n "$NPROC_PER_NODE" ]; then
+    WORLD_SIZE=$ARG_WORLD_SIZE
+    NPROC_PER_NODE=$ARG_NPROC_PER_NODE
+fi
+if [ ! -n "$MASTER_ADDR" ] || [ ! -n "$MASTER_PORT" ] || [ ! -n "$RANK" ]; then
+    MASTER_ADDR=$ARG_MASTER_ADDR
+    MASTER_PORT=$ARG_MASTER_PORT
+    RANK=$ARG_RANK
+fi
+
+echo "WORLD_SIZE: $WORLD_SIZE"
+echo "NPROC_PER_NODE: $NPROC_PER_NODE"
+
+# Training Arguments
+GLOBAL_BATCH_SIZE=256
+LOCAL_BATCH_SIZE=8
+GRADIENT_ACCUMULATION_STEPS=$[$GLOBAL_BATCH_SIZE/($WORLD_SIZE*$NPROC_PER_NODE*$LOCAL_BATCH_SIZE)]
+echo $GRADIENT_ACCUMULATION_STEPS
+
+# Log Arguments
+export TRANSFORMERS_OFFLINE=1
+export WANDB_PROJECT=videollama2qwen2_siglip
+RUN_NAME=vllava_settings
+DATA_DIR=datasets
+OUTP_DIR=work_dirs
+
+torchrun --nnodes $WORLD_SIZE \
+    --nproc_per_node $NPROC_PER_NODE  \
+    --master_addr=$MASTER_ADDR \
+    --master_port=$MASTER_PORT \
+    --node_rank $RANK \
+    videollama2/train_flash_attn.py \
+    --deepspeed scripts/zero3.json \
+    --model_type videollama2_qwen2 \
+    --model_path Qwen/Qwen2-7B-Instruct \
+    --vision_tower google/siglip-so400m-patch14-384 \
+    --mm_projector_type stc_connector_v35 \
+    --tune_mm_mlp_adapter True \
+    --data_path   ${DATA_DIR}/videollava_pt/valley_llavaimage.json \
+    --data_folder ${DATA_DIR}/videollava_pt/ \
+    --mm_vision_select_layer -2 \
+    --num_frames 8 \
+    --bf16 True \
+    --tf32 True \
+    --fp16 False \
+    --output_dir ${OUTP_DIR}/${WANDB_PROJECT}/pretrain_${RUN_NAME} \
+    --num_train_epochs 1 \
+    --per_device_train_batch_size $LOCAL_BATCH_SIZE \
+    --per_device_eval_batch_size 4 \
+    --gradient_accumulation_steps $GRADIENT_ACCUMULATION_STEPS \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 500 \
+    --save_total_limit 99 \
+    --learning_rate 1e-3 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --model_max_length 2048 \
+    --gradient_checkpointing True \
+    --dataloader_num_workers 4 \
+    --lazy_preprocess True \
+    --report_to tensorboard \
+    --run_name $RUN_NAME \

scripts/vllava/finetune.sh

@@ -0,0 +1,74 @@
+#!/bin/bash
+
+# Environment Variables
+ARG_WORLD_SIZE=${1:-1}
+ARG_NPROC_PER_NODE=${2:-8}
+ARG_MASTER_ADDR="127.0.0.1"
+ARG_MASTER_PORT=16666
+ARG_RANK=0
+
+# Multiple conditions
+if [ ! -n "$WORLD_SIZE" ] || [ ! -n "$NPROC_PER_NODE" ]; then
+    WORLD_SIZE=$ARG_WORLD_SIZE
+    NPROC_PER_NODE=$ARG_NPROC_PER_NODE
+fi
+if [ ! -n "$MASTER_ADDR" ] || [ ! -n "$MASTER_PORT" ] || [ ! -n "$RANK" ]; then
+    MASTER_ADDR=$ARG_MASTER_ADDR
+    MASTER_PORT=$ARG_MASTER_PORT
+    RANK=$ARG_RANK
+fi
+
+echo "WORLD_SIZE: $WORLD_SIZE"
+echo "NPROC_PER_NODE: $NPROC_PER_NODE"
+
+# Training Arguments
+GLOBAL_BATCH_SIZE=128
+LOCAL_BATCH_SIZE=4
+GRADIENT_ACCUMULATION_STEPS=$[$GLOBAL_BATCH_SIZE/($WORLD_SIZE*$NPROC_PER_NODE*$LOCAL_BATCH_SIZE)]
+
+# Log Arguments
+export TRANSFORMERS_OFFLINE=1
+export WANDB_PROJECT=videollama2
+RUN_NAME=vllava_settings
+DATA_DIR=datasets
+OUTP_DIR=work_dirs
+
+torchrun --nnodes $WORLD_SIZE \
+    --nproc_per_node $NPROC_PER_NODE \
+    --master_addr=$MASTER_ADDR \
+    --master_port=$MASTER_PORT \
+    --node_rank $RANK \
+    videollama2/train_flash_attn.py \
+    --deepspeed scripts/zero3.json \
+    --model_type videollama2 \
+    --model_path mistralai/Mistral-7B-Instruct-v0.2 \
+    --vision_tower openai/clip-vit-large-patch14-336 \
+    --mm_projector_type stc_connector_v35 \
+    --pretrain_mm_mlp_adapter ${OUTP_DIR}/${WANDB_PROJECT}/pretrain_${RUN_NAME}/mm_projector.bin \
+    --data_path   ${DATA_DIR}/videollava_sft/videochatgpt_llavaimage_tune.json \
+    --data_folder ${DATA_DIR}/videollava_sft/ \
+    --mm_vision_select_layer -2 \
+    --image_aspect_ratio pad \
+    --num_frames 8 \
+    --bf16 True \
+    --tf32 True \
+    --fp16 False \
+    --output_dir ${OUTP_DIR}/${WANDB_PROJECT}/finetune_${RUN_NAME} \
+    --num_train_epochs 1 \
+    --per_device_train_batch_size $LOCAL_BATCH_SIZE \
+    --per_device_eval_batch_size 4 \
+    --gradient_accumulation_steps $GRADIENT_ACCUMULATION_STEPS \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 500 \
+    --save_total_limit 99 \
+    --learning_rate 2e-5 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --model_max_length 2048 \
+    --gradient_checkpointing True \
+    --dataloader_num_workers 4 \
+    --report_to tensorboard \
+    --run_name $RUN_NAME \

scripts/vllava/pretrain.sh

@@ -0,0 +1,74 @@
+#!/bin/bash
+
+# Environment Variables
+ARG_WORLD_SIZE=${1:-1}
+ARG_NPROC_PER_NODE=${2:-8}
+ARG_MASTER_ADDR="127.0.0.1"
+ARG_MASTER_PORT=16666
+ARG_RANK=0
+
+# Multiple conditions
+if [ ! -n "$WORLD_SIZE" ] || [ ! -n "$NPROC_PER_NODE" ]; then
+    WORLD_SIZE=$ARG_WORLD_SIZE
+    NPROC_PER_NODE=$ARG_NPROC_PER_NODE
+fi
+if [ ! -n "$MASTER_ADDR" ] || [ ! -n "$MASTER_PORT" ] || [ ! -n "$RANK" ]; then
+    MASTER_ADDR=$ARG_MASTER_ADDR
+    MASTER_PORT=$ARG_MASTER_PORT
+    RANK=$ARG_RANK
+fi
+
+echo "WORLD_SIZE: $WORLD_SIZE"
+echo "NPROC_PER_NODE: $NPROC_PER_NODE"
+
+# Training Arguments
+GLOBAL_BATCH_SIZE=256
+LOCAL_BATCH_SIZE=8
+GRADIENT_ACCUMULATION_STEPS=$[$GLOBAL_BATCH_SIZE/($WORLD_SIZE*$NPROC_PER_NODE*$LOCAL_BATCH_SIZE)]
+
+# Log Arguments
+export TRANSFORMERS_OFFLINE=1
+export WANDB_PROJECT=videollama2
+RUN_NAME=vllava_settings
+DATA_DIR=datasets
+OUTP_DIR=work_dirs
+
+torchrun --nnodes $WORLD_SIZE \
+    --nproc_per_node $NPROC_PER_NODE  \
+    --master_addr=$MASTER_ADDR \
+    --master_port=$MASTER_PORT \
+    --node_rank $RANK \
+    videollama2/train_flash_attn.py \
+    --deepspeed scripts/zero3.json \
+    --model_type videollama2 \
+    --model_path mistralai/Mistral-7B-Instruct-v0.2 \
+    --vision_tower openai/clip-vit-large-patch14-336 \
+    --mm_projector_type stc_connector_v35 \
+    --tune_mm_mlp_adapter True \
+    --data_path   ${DATA_DIR}/videollava_pt/valley_llavaimage.json \
+    --data_folder ${DATA_DIR}/videollava_pt/ \
+    --mm_vision_select_layer -2 \
+    --num_frames 8 \
+    --bf16 True \
+    --tf32 True \
+    --fp16 False \
+    --output_dir ${OUTP_DIR}/${WANDB_PROJECT}/pretrain_${RUN_NAME} \
+    --num_train_epochs 1 \
+    --per_device_train_batch_size $LOCAL_BATCH_SIZE \
+    --per_device_eval_batch_size 4 \
+    --gradient_accumulation_steps $GRADIENT_ACCUMULATION_STEPS \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 500 \
+    --save_total_limit 99 \
+    --learning_rate 1e-3 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --model_max_length 2048 \
+    --gradient_checkpointing True \
+    --dataloader_num_workers 4 \
+    --lazy_preprocess True \
+    --report_to tensorboard \
+    --run_name $RUN_NAME \