chavinlo

Results 78 comments of chavinlo

Stable diffusion examples generates bad images

Hello, I am having a similar issue. I am using an A100 as reported by nvidia-smi: ``` Fri Dec 23 21:47:17 2022 +-----------------------------------------------------------------------------+ | NVIDIA-SMI 520.61.05 Driver Version: 520.61.05 CUDA...

ValueError: ConstantTensor's maximum size is not equal to len(data)!

Looks like so. Reverting to commit 3625c3056041963926ba455347d9091fb891b872 fixes the issue.

Train to be a perfect loop

https://github.com/chavinlo/musicgen_trainer done

LoRAs possible for fast and lightweight fine-tuning?

you first have to figure out how to train

Example use case of `compute_predictions`?

*I don't know much about lm training, so excuse me if I am missing something obvious

How do I run this?

make_a_video.py only contains the resources needed for make-a-video, a make-a-video model doesn't even exists (in the public) yet.

I combined your code with diffusers stable diffusion and trained a model

Amazing job.

I combined your code with diffusers stable diffusion and trained a model

> If you need more explanations, I could also share my train_dreambooth.py Could it be possible to do so?

I combined your code with diffusers stable diffusion and trained a model

> @chavinlo I dropped my messy script at https://gist.github.com/lxj616/5134368f44aca837304530695ee100ea > > But I bet it would be quicker if you modify the original train_dreambooth.py from diffusers than to debug mine,...

I combined your code with diffusers stable diffusion and trained a model

I've read your blog about VRAM limitations. If you need more compute, I can give you an A100 to experiment.