v1.0.0

tobias-kirschstein/ggheadv1.0.0Mar 17, 2024by FrankLeeeee

AI Summary

This initial release establishes the Open-Sora project as an accessible platform for efficient video generation, providing the core source code and optimized infrastructure. It includes essential training and inference pipelines optimized with ColossalAI and flash attention, along with initial model weights.

Key Highlights

Introduction of the Open-Sora initiative for accessible video generation
Release of source code for the training and inference pipeline
System optimization using ColossalAI and flash attention
Availability of initial model weights trained on specific resolutions

New Features

Data processing source code
Training and inference pipeline source code
ColossalAI and flash attention optimization
Initial model weights

Full Release Notes

We present Open-Sora, an initiative dedicated to efficiently produce high-quality video and make the model, tools and contents accessible to all. By embracing open-source principles, Open-Sora not only democratizes access to advanced video generation techniques, but also offers a streamlined and user-friendly platform that simplifies the complexities of video production. With Open-Sora, we aim to inspire innovation, creativity, and inclusivity in the realm of content creation.

In this release, we have included
- source code for data processing
- source code for sora-like video generation training and inference pipeline
- system optimization powered by ColossalAI and flash attention
- model weights trained for 16×256×256 and 16×512x512 resolutions, available on Hugging Face