# Flax/JAX community week ��
Welcome to the Flax/JAX community week! The goal of this week is to make compute-intensive NLP and CV projects (like pre-training BERT, GPT2, CLIP, ViT)
practicable for a wider audience of engineers and researchers.
To do so, we will try to teach **you** how to effectively use JAX/Flax on TPU and help you to complete a fun NLP and/or CV project in JAX/Flax during the community week.
Free access to a TPUv3-8 will kindly be provided by the Google Cloud team!
In this document, we list all the important information that you will need during the Flax/JAX community week.
Don't forget to sign up [here](https://forms.gle/tVGPhjKXyEsSgUcs8)!
## Table of Contents
- [Organization](#organization)
- [Important dates](#important-dates)
- [Communication](#communication)
- [Projects](#projects)
- [How to propose](#how-to-propose-a-project)
- [How to form a team](#how-to-form-a-team-around-a-project)
- [Tips & Tricks for project](#tips-on-how-to-organize-the-project)
- [How to install flax, jax, optax, transformers, datasets](#how-to-install-relevant-libraries)
- [Quickstart Flax/JAX](#quickstart-flax-and-jax)
- [Quickstart Flax/JAX in �� Transformers](#quickstart-flax-and-jax-in-transformers)
- [Flax design philosophy in �� Transformers](#flax-design-philosophy-in-transformers)
- [How to use flax models & scripts](#how-to-use-flax-models-and-example-scripts)
- [Talks](#talks)
- [How to use the �� Hub for training](#how-to-use-the-hub-for-collaboration)
- [How to setup TPU VM](#how-to-setup-tpu-vm)
- [How to build a demo](#how-to-build-a-demo)
- [Using the Hugging Face Widgets](#using-the-hugging-face-widgets)
- [Using a Streamlit demo](#using-a-streamlit-demo)
- [Using a Gradio demo](#using-a-gradio-demo)
- [Project evaluation](#project-evaluation)
- [General Tips & Tricks](#general-tips-and-tricks)
- [FAQ](#faq)
## Organization
Participants can propose ideas for an interesting NLP and/or CV project. Teams of 3 to 5 will then be formed around the most promising and interesting projects. Make sure to read through the [Projects](#projects) section on how to propose projects, comment on other participants' project ideas, and create a team.
To help each team successfully finish their project, we have organized talks by leading scientists and engineers from Google, Hugging Face, and the open-source NLP & CV community. The talks will take place before the community week from June 30th to July 2nd. Make sure to attend the talks to get the most out of your participation! Check out the [Talks](#talks) section to get an overview of the talks, including the speaker and the time of the talk.
Each team is then given **free access to a TPUv3-8 VM** from July 7th to July 14th. In addition, we will provide training examples in JAX/Flax for a variety of NLP and Vision models to kick-start your project. During the week, we'll make sure to answer any questions you might have about JAX/Flax and Transformers and help each team as much as possible to complete their project!
At the end of the community week, each team should submit a demo of their project. All demonstrations will be evaluated by a jury and the top-3 demos will be awarded a prize. Check out the [How to submit a demo](#how-to-submit-a-demo) section for more information and suggestions on how to submit your project.
## Important dates
- **23.06.** Official announcement of the community week. Make sure to sign-up in [this google form](https://forms.gle/tVGPhjKXyEsSgUcs8).
- **23.06. - 30.06.** Participants will be added to an internal Slack channel. Project ideas can be proposed here and groups of 3-5 are formed. Read this document for more information.
- **30.06.** Release of all relevant training scripts in JAX/Flax as well as other documents on how to set up a TPU, how to use the training scripts, how to submit a demo, tips & tricks for JAX/Flax, tips & tricks for efficient use of the hub.
- **30.06. - 2.07.** Talks about JAX/Flax, TPU, Transformers, Computer Vision & NLP will be held.
- **7.07.** Start of the community week! Access to TPUv3-8 will be given to each team.
- **7.07. - 14.07.** The Hugging Face & JAX/Flax & Cloud team will be available for any questions, problems the teams might run into.
- **15.07.** Access to TPU is deactivated and community week officially ends.
- **16.07.** Deadline for each team to submit a demo.
## Communication
All important communication will take place in an internal Slack channel, called `#flax-jax-community-week`.
Important announcements of the Hugging Face, Flax/JAX, and Google Cloud team will be posted there.
Such announcements include general information about the community week (Dates, Rules, ...), release of relevant training scripts (Flax/JAX example scripts for NLP and Vision), release of other important documents (How to access the TPU), etc.
The Slack channel will also be the central place for participants to post about their results, share their learning experiences, ask questions, etc.
For issues with Flax/JAX, Transformers, Datasets or for questions that are specific to your project we would be **very happy** if you could use the following public repositories and forums:
- Flax: [Issues](https://github.com/google/flax/issues), [Questions](https://github.com/google/flax/discussions)
- JAX: [Issues](https://github.com/google/jax/issues), [Questions](https://github.com/google/jax/discussions)
- �� Transformers: [Issues](https://github.com/huggingface/transformers/issues), [Questions](https://discuss.huggingface.co/c/transformers/9)
- �� Datasets: [Issues](https://github.com/huggingface/datasets/issues), [Questions](https://discuss.huggingface.co/c/datasets/10)
- Project specific questions: [Forum](https://discuss.huggingface.co/c/flax-jax-projects/22)
- TPU related questions: [TODO]()
Please do **not** post the complete issue/project-specific question in the Slack channel, but instead a link to your issue/question that we will try to answer as soon as possible.
This way, we make sure that the everybody in the community can benefit from your questions - even after the community week - and that the same question is not answered twice.
To be invited to the Slack channel, please make sure you have signed up [on the Google form](https://forms.gle/tVGPhjKXyEsSgUcs8).
**Note**: If you have signed up on the google form, but you are not in the Slack channel, please leave a message on [(TODO) the official forum announcement]( ) and ping `@Suzana` and `@patrickvonplaten`.
## Projects
During the first week after the community week announcement, **23.06. - 30.06.**, teams will be formed around the most promising and interesting project ideas. Each team can consist of 2 to 10 participants. Projects can be accessed [here](https://discuss.huggingface.co/c/flax-jax-projects/22).
All officially defined projects can be seen [here](https://docs.google.com/spreadsheets/d/1GpHebL7qrwJOc9olTpIPgjf8vOS0jNb6zR_B8x_Jtik/edit?usp=sharing).
### How to propose a project
Some default project ideas are given by the organizers. **However, we strongly encourage participants to submit their own project ideas!**
Check out the [HOW_TO_PROPOSE_PROJECT.md](https://github.com/huggingface/transformers/tree/main/examples/research_projects/jax-projects/HOW_TO_PROPOSE_PROJECT.md) for more information on how to propose a new project.
### How to form a team around a project
You can check out all existing projects ideas on the forum under [Flax/JAX projects category](https://discuss.huggingface.co/c/flax-jax-projects/22).
Make sure to quickly check out each project idea and leave a �わ� if you like an idea.
Feel free to leave comments, suggestions for improvement, or questions about more details directly on the discussion thread.
If you have found the project that you �わ� the most, leave a message "I would like to join this project" on the discussion thread.
We strongly advise you to also shortly state who
没有合适的资源?快使用搜索试试~ 我知道了~
Transformers:适用于 Pytorch、TensorFlow 和 JAX 的最先进的机器学习
共2000个文件
py:1850个
md:103个
json:17个
需积分: 5 0 下载量 21 浏览量
2024-09-14
11:18:31
上传
评论
收藏 21.92MB ZIP 举报
温馨提示
Transformers 提供数千个预训练模型,用于执行文本、视觉和音频等不同模态的任务。 这些模型可以应用于: 文本,用于文本分类、信息提取、问答、摘要、翻译和文本生成等任务,支持 100 多种语言。 图像,用于图像分类、对象检测和分割等任务。 音频,用于语音识别和音频分类等任务。 Transformer 模型还可以结合几种模式执行任务,例如表格问答、光学字符识别、从扫描文档中提取信息、视频分类和视觉问答。 Transformers 提供 API,可快速下载和使用针对给定文本的预训练模型,在您自己的数据集上对其进行微调,然后在我们的模型中心与社区分享。同时,定义架构的每个 Python 模块都是完全独立的,可以进行修改以实现快速的研究实验。 Transformers 由三个最受欢迎的深度学习库(Jax、PyTorch和TensorFlow)提供支持,它们之间无缝集成。使用其中一个库训练模型很简单,然后再使用另一个库加载模型进行推理。
资源推荐
资源详情
资源评论
收起资源包目录
Transformers:适用于 Pytorch、TensorFlow 和 JAX 的最先进的机器学习 (2000个子文件)
wkv_op.cpp 4KB
fast_lsh_cumulation_torch.cpp 3KB
torch_extension.cpp 1KB
ms_deform_attn_cpu.cpp 1KB
ms_deform_attn_cpu.cpp 1KB
vision.cpp 798B
vision.cpp 798B
fast_lsh_cumulation_cuda.h 5KB
common_cuda_device.h 2KB
ms_deform_attn.h 2KB
ms_deform_attn.h 2KB
ms_deform_attn_cuda.h 2KB
cuda_kernel.h 2KB
fast_lsh_cumulation.h 2KB
ms_deform_attn_cuda.h 1KB
ms_deform_attn_cpu.h 1KB
ms_deform_attn_cpu.h 1KB
cuda_launch.h 707B
common.h 273B
common_cuda.h 258B
tiny_model_summary.json 202KB
test.json 27KB
val.json 21KB
sample.json 17KB
train.json 11KB
sample.json 3KB
sample.json 3KB
ds_config_zero3.json 1KB
sample.json 1KB
ds_config_zero2.json 1KB
add_distilbert_like_config.json 504B
vocab.json 228B
dummy_feature_extractor_config.json 101B
preprocessor_config.json 100B
test_entity_vocab.json 76B
vit_feature_extractor.json 72B
dummy-config.json 29B
README.md 82KB
README.md 41KB
awesome-transformers.md 38KB
README.md 37KB
README.md 37KB
README.md 23KB
README.md 19KB
README.md 19KB
README.md 19KB
README.md 16KB
README.md 16KB
README.md 15KB
README.md 15KB
README.md 12KB
README.md 11KB
README.md 11KB
README.md 11KB
README.md 10KB
README.md 10KB
README.md 10KB
README.md 10KB
README.md 10KB
README.md 10KB
README.md 10KB
README.md 9KB
README.md 9KB
README.md 9KB
README.md 9KB
README.md 8KB
README.md 8KB
README.md 8KB
README.md 7KB
README.md 7KB
README.md 7KB
README.md 7KB
README.md 7KB
README.md 7KB
README.md 6KB
README.md 6KB
README.md 6KB
README.md 6KB
README.md 6KB
README.md 5KB
README.md 5KB
README.md 5KB
README.md 5KB
README.md 4KB
README.md 4KB
README.md 4KB
README.md 4KB
README.md 4KB
README.md 3KB
README.md 3KB
README.md 3KB
README.md 3KB
README.md 3KB
README.md 3KB
README.md 3KB
PULL_REQUEST_TEMPLATE.md 3KB
README.md 3KB
README.md 3KB
README.md 3KB
README.md 3KB
共 2000 条
- 1
- 2
- 3
- 4
- 5
- 6
- 20
资源评论
余十步
- 粉丝: 1675
- 资源: 172
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
最新资源
- Adams多体动力学仿真 Cruise 动力性,经济性仿真 动力系统参数匹配 纯电动汽车,混合动力汽车,双电机汽车 simuli
- 基于Java语言的SpringCloud微服务架构设计源码
- 基于Python的多种系统辨识算法设计源码
- 基于C++的北京六部工坊机器人仿真开源设计源码
- Qt实现的包含图片显示功能、自适应缩放、背景图片、画roi工具 不依赖其他库纯Qt实现 在图片上可以画矩形、矩形旋转、圆形、
- 基于Springboot+Vue前后端分离技术的课程精品网站设计源码
- 基于SpringBoot和Vue的民宿在线预定平台设计源码
- 正余弦优化(SCA)算法+任务分配的多智能体路径规划算法 matlab代码 显示方式为动态显示 新型群智能算法 多智能体
- 基于Java语言的广东海洋大学体育管理系统后端设计源码
- 基于Java实现的boardRoom会议室控制设计源码
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功