sashasubbbb
sashasubbbb
I'm getting same error `AttributeError: 'NoneType' object has no attribute 'filename'` When using a sdxl lora on sdxl model EDIT: Nevermind, seems lora i'm using is broken.
I agree, this is a much needed extenstion.
Same on 2070 super 8gb. Cannot run hires fix without xformers on. I'm on ` python: 3.10.6 • torch: 2.1.0.dev20230317+cu118 • xformers: N/A • gradio: 3.16.2 • commit: [a9fed7c3]` Tried...
Yeah, this solution worked ok but without --cache_latents on. When i use caching, it seems to ignore dataset_repeats flag. Is there any other way to set multiple repeats of dataset...
For 8gb VRAM i was able to run LLaMA-7B in 8bit mode with --gpu-memory 7 (but it did not really help with anything when generating gets me OOM 20MB). Anything...
@machachu56 thank you for your notebook. Is there any way to specify exact total epoch count and save each n epoch parameters?
Any ETA for refiner?
Yeah, seems pretty useful.