Hugging Face's logo

Spaces:
hugginglearners
/
README
Running

App Files Community
README / README.md
espejelomar's picture
espejelomar
Update README.md
26bc153
preview code
|
raw
history blame
19.7 kB
metadata
title: README
emoji: 💻
colorFrom: indigo
colorTo: gray
sdk: static
pinned: false

👋👋👋👋👋 Welcome to the fastai X Hugging Face Group 🥳🥳🥳🥳🥳

Join the Group by clicking here

Live Group Leaderboard

Hugging Learners

Few have done as much as the fast.ai ecosystem to make Deep Learning accessible. Let's make exclusivity in access to Machine Learning, including pre-trained models, a thing of the past, and let's push this unique field even further.


Goal

1 month (June 15 to July 15) group to share Vision and Text pre-trained fastai Learners with the community for further community usage and reproducibility.


Why? Community, collaboration, and reproducibility

We believe in openly sharing knowledge and resources to democratize AI for all. At Hugging Face, we encourage all practitioners who train models to contribute by sharing them with the community. Even when trained on particular data sets, sharing Learners will help others save time and computing resources, and give them access to valuable trained artifacts. In turn, you can benefit from the work that others have done. Additionally, shared Learners can be replicated by other community members through, for example, the inference API or repository cloning.


What should I do?

This tutorial shows how to share and load Learners (including those created with blurr) to and from the Hugging Face Hub.

  • Say “Hi!” in the hf-fastai channel in the fastai Discord server (kudos Wayde Gilliam). Servers are places where a community can interact via channels for specific topics.
  • Join here to the “fastai X Hugging Face Group 2022 (hugginglearners)” organization in the Hub (so that you can add your Learner to the organization).
  • Train a vision or text fastai Learner. You can use an already pre-trained model from the Hugging Face Hub.
  • Share a trained fastai Learner with the community through the Hub. You can use the push_to_hub_fastai function. Remember to describe your model in the model card.
  • Create a Gradio demo to showcase your model. Here is a tutorial in the Hugging Face course, and the next iteration of the fastai course will include a section on sharing with Gradio.

SWAG and prizes

  • Participants that share at least two Learners with their respective Gradio Space will get a voucher to acquire the official Study Group t-shirt in the Hugging Face merch store. The graphic design will be defined soon.
  • All participants: fastai x Hugging Face event badge on HF.
  • Top 10 Gradio Spaces based on likes: Hugging Face PRO subscription for 1 month

What is Hugging Face Spaces?

Spaces are a simple way to host ML demo apps directly on your profile or your organization’s profile on Hugging Face. This allows you to create your ML portfolio, showcase your projects at conferences or to stakeholders, and work collaboratively with other people in the ML ecosystem. Learn more about spaces here.


We are happy to invite you to the Gradio Blocks Party - a community event in which we will create <b>interactive demos</b> for state-of-the-art machine learning models. Demos are powerful because they allow anyone — not just ML engineers — to try out models in the browser, give feedback on predictions, identify trustworthy models. The event will take place from <b>May 17th to 31st</b>. We will be organizing this event on <a href="https://huggingface.co/Gradio-Blocks" target="_blank" style="text-decoration: underline">Hugging Face</a> and the <a href="https://discord.com/invite/feTf9x3ZSB" target="_blank" style="text-decoration: underline">Hugging Face discord server</a>, check the gradio channel for staying up to date with the event. Prizes will be given at the end of the event, see the <a href="#Prizes" target="_blank" style="text-decoration: underline">Prizes section</a> </p><br />

We will be building demos using the new Gradio Blocks API. Blocks allows you to build web-based demos in a flexible way using the Gradio library. Gradio is a popular choice for building demos for machine learning models, as it allows you to create web-based UIs all in Python. For example, here is a UI for Dall-E Mini using Gradio Blocks: 



  <h3  class="my-8 lg:col-span-2" style="font-size:20px; font-weight:bold">More About Gradio</h3>

Gradio is a Python library that allows you to quickly build web-based machine learning demos, data science dashboards, or other kinds of web apps, entirely in Python. These web apps can be launched from wherever you use Python (jupyter notebooks, colab notebooks, Python terminal, etc.) and shared with anyone instantly using Gradio's auto-generated share links. To learn more about Gradio see the Getting Started Guide: https://gradio.app/getting_started/ and the new Course on Huggingface about Gradio: Gradio Course.


Gradio can be installed via pip and comes preinstalled in Hugging Face Spaces, the latest version of Gradio can be set in the README in spaces by setting the sdk_version, for example sdk_version: 3.0.2


To install gradio locally, simply run: pip install gradio


Gradio Blocks

gradio.Blocks is a low-level API that allows you to have full control over the data flows and layout of your application. You can build very complex, multi-step applications using Blocks. If you have already used gradio.Interface, you know that you can easily create fully-fledged machine learning demos with just a few lines of code. The Interface API is very convenient but in some cases may not be sufficiently flexible for your needs. For example, you might want to:

  • Group together related demos as multiple tabs in one web app
  • Change the layout of your demo instead of just having all of the inputs on the left and outputs on the right
  • Have multi-step interfaces, in which the output of one model becomes the input to the next model, or have more flexible data flows in general
  • Change a component's properties (for example, the choices in a Dropdown) or its visibilty based on user input

to learn more about Blocks see the guide https://www.gradio.app/introduction_to_blocks/


What is Hugging Face Spaces?

Spaces are a simple way to host ML demo apps directly on your profile or your organization’s profile on Hugging Face. This allows you to create your ML portfolio, showcase your projects at conferences or to stakeholders, and work collaboratively with other people in the ML ecosystem. Learn more about spaces here.


How Does Gradio and Hugging Face work together?

Hugging Face Spaces is a free hosting option for Gradio demos. Spaces comes with 3 SDK options: Gradio, Streamlit and Static HTML demos. Spaces can be public or private and the workflow is similar to github repos. There are over 2000+ Gradio spaces currently on Hugging Face. Learn more about spaces here: https://huggingface.co/docs/hub/spaces


Event Plan


Some examples using Blocks

mindseye-lite

To participate in the event


Potential ideas for creating spaces:


Prizes


Other Potential Prizes

  • Staff Picks
  • Most liked Spaces
  • Community Pick (voting)
  • Most Creative Space (voting)
  • Most Educational Space (voting)
  • CEO's pick (one prize for a particularly impactful demo), picked by @clem
  • CTO's pick (one prize for a particularly technically impressive demo), picked by @julien

Creating a Gradio demo on Hugging Face Spaces

Once a model has been picked from the choices above, you can share a model in a Space using Gradio. Read more about how to add Gradio spaces: https://huggingface.co/blog/gradio-spaces

Steps to add Gradio Spaces to the Gradio Blocks Party org

  • Create a account on Hugging Face
  • Join the Gradio Blocks Party Organization by clicking "Join Organization" button in the organization page or using the shared link above
  • Once your request is approved, add your space using the Gradio SDK and share the link with the community!

LeaderBoard for Most Popular Blocks Event Spaces

See the Live Blocks Party Leaderboard