https://github.com/p1atdev/vision-pt

https://github.com/p1atdev/vision-pt

Last synced: 15 days ago
JSON representation

• Host: GitHub
• URL: https://github.com/p1atdev/vision-pt
• Owner: p1atdev
• License: apache-2.0
• Created: 2025-12-03T11:19:06.000Z (7 months ago)
• Default Branch: main
• Last Pushed: 2026-01-13T07:33:07.000Z (5 months ago)
• Last Synced: 2026-01-13T08:43:38.244Z (5 months ago)
• Language: Python
• Size: 3.23 MB
• Stars: 5
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Vision model pretraining scripts

WIP

## Features

- NF4 model loading 

- QLoRA with bitsandbytes

- [Flash Attention](https://github.com/Dao-AILab/flash-attention) support

- Aspect ratio bucketing

## Models

- [x] JiT

## References

- https://github.com/kohya-ss/sd-scripts

  - Heavily inspired by this repository.

- https://github.com/cloneofsimo/minRF

- https://github.com/huggingface/diffusers

- https://github.com/Stability-AI/sd3-ref

- https://github.com/black-forest-labs/flux

- https://github.com/Alpha-VLLM/Lumina-Image-2.0

  - Model implementation

- https://github.com/microsoft/LoRA

- https://github.com/huggingface/peft

  - Peft logic and implementation

- https://github.com/huggingface/diffusers

- https://github.com/Lightning-AI/pytorch-lightning

- https://github.com/ostris/ai-toolkit

  - Traning cycle and API design

- https://github.com/NovelAI/novelai-aspect-ratio-bucketing

  - Aspect ratio bucketing

- https://github.com/bitsandbytes-foundation/bitsandbytes

  - Quantization