Linaqruf / kohya-trainer

Adapted from https://note.com/kohya_ss/n/nbf7ce8d80f29 for easier cloning
Apache License 2.0
1.86k stars 309 forks source link

Training ok, but Lora doesn't work #313

Open ermanitu opened 1 year ago

ermanitu commented 1 year ago

I trained several models with colab version of Lora trainer XL. All seems go ok, but the trained model doesn't work in automatic1111.

No problem with basic Lora Trainer (SD2 and SD1). I tried a lot of different parameters configuration without success.

Any suggestion? Someone has used Lora Trainer XL in colab and run the model in automatic1111?