title: Feel emoji: 🚀 colorFrom: blue colorTo: gray sdk: gradio sdk_version: 5.10.0 app_file: app/app.py pinned: false

Feel

This is a project to create a continuous training application.

Platform being developed at MIT in collaboration with HuggingFace. Aimed at improving performance of existing Large Language Models through real-time human feedback loop.

This repository hosts the development of an automated RLHF platform for Hugging Face, where the community can provide real-time feedback on language models. The feedback is automatically integrated into an RLHF pipeline to continuously fine-tune and improve the models.

What is Feel?

A community-driven project to improve Multilingual Vision-Language Models (VLMs). Leverages feedback from users and automated RLHF pipelines to continuously improve model performance.

Why Feel?

Feel is a platform that enables the community to provide real-time feedback on language models. The feedback is automatically integrated into an RLHF pipeline to continuously fine-tune and improve the models.

Repository Structure

The repository is organized as follows:

ml/                # Directory for machine learning code
├── README.md      # Dataset schema and project structure
├── data/          # Directory for dataset files
├── models/        # Directory for model files
app/               # Directory for application code
├── app.py         # Main application file

Installation

The repository uses uv for managing virtual environments. To install uv, go here. Create a virtual environment.

uv venv --python 3.11

Activate the virtual environment

source .venv/bin/activate

Sync the dependencies

uv sync --all-groups

To install the required dependencies, run the following commands:

ML Dependencies

uv sync --group ml

App Dependencies

uv sync --group app

Sync with Hugging Face Space

First, add the remote

git remote add hf https://huggingface.co/spaces/feel-fl/open-human-feedback-chat

Push to the remote branch

git push hf main

feel

Commits

Refactor language handling in app.py to use a dictionary for system messages, enhancing support for multiple languages and simplifying message retrieval.

Add debugging breakpoint in submit_conversation function to inspect conversation processing

Refactor URL generation in format_history_as_messages to use consistent variable naming for file paths, improving code clarity and maintainability.

Fix URL generation in format_history_as_messages to correctly format file paths by removing unnecessary encoding in the URL string.

Enhance URL generation in format_history_as_messages by adding print statement for debugging

Update image URL handling in format_history_as_messages function to support SPACE_HOST environment variable

README