The training script for lora and full finetune is already now.
#25
by duongve - opened
Based on the contributions of @Bluvoll and comfyui, I have completed the implementation for both LoRA and full fine-tuning. You can check my pull request here: https://github.com/tdrussell/diffusion-pipe/pull/505. Here are the results.
This is the base model:
This is the finetune model:
Note: The training script for kohy, i will release soon when i complete it :D
duongve changed discussion status to closed
duongve changed discussion status to open
I wonder, why dual text encoders? Anima uses just qwen, right?
duongve changed discussion status to closed