This commit is contained in:
Darren Lau
2025-03-06 16:48:05 +08:00
parent 040c04f17e
commit 190df71e6d

View File

@@ -1,3 +1,13 @@
# Custom trainer dedicated for AstolfoXL #
- Rewritten `sdxl_train.py` to gather many features from stalled PR and exclusively in other scripts. [The PR](https://github.com/kohya-ss/sd-scripts/pull/1950/).
- Many added features are *obvious and fundamental* for large scale finetuning: Multi node multi GPU (scalable k8s scripts not included), resume from steps, and common enhancing techniques.
- Code has been refractored as `train_native.py`, which would replace `fine_tune.py` which matches the code pattern from `train_network.py`. *However there are many arch specific problem to be solved.*
------
This repository contains training, generation and utility scripts for Stable Diffusion.
## FLUX.1 and SD3 training (WIP)