Skip to content

README.md

Latest commit

 

History

History
60 lines (41 loc) · 1.12 KB

README.md

File metadata and controls

60 lines (41 loc) · 1.12 KB

FluxGym Trainer

A Modal-based application for training Stable Diffusion models using FluxGym interface.

Features

  • Runs on Modal cloud platform
  • Uses NVIDIA CUDA for GPU acceleration
  • Integrates with HuggingFace
  • Web-based UI powered by Gradio
  • Supports A100 GPU for training

Prerequisites

  • Modal account
  • HuggingFace account and API token
  • Python 3.11+

Setup

  1. Install Modal CLI:
pip install modal
  1. Configure Modal:
modal token new
  1. Set up HuggingFace secret in Modal:
modal secret create huggingface-secret HUGGINGFACE_SECRET=your_token_here

Running the Application

  1. To run on modal:
modal serve app.py

Environment

  • CUDA Version: 12.4.0
  • Base OS: Ubuntu 22.04
  • Python Version: 3.11
  • GPU: NVIDIA A100

Architecture

The application uses Modal's serverless infrastructure to run a Gradio-based web interface for training FLUX models. It utilizes persistent volume storage for outputs and integrates with HuggingFace for model management.

License

MIT License

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.