search

Stability-AI / stable-audio-tools

Generative models for conditional audio generation
MIT License
2.73k stars 258 forks source link

Finetuning stable-audio-open 1.0 - Model Weights + Configs #138

Closed cvillela closed 2 months ago

cvillela commented 3 months ago

Hey there!

First of all congratulations on the model, and thanks for open sourcing!

I wanted to finetune the Stable Audio Open model, as oposed to Stable Audio 1 or 2, as the results on the paper look so much more promising than the previous models.

There is already an issue #112 requesting the open weights for the model.

I was also interested in the model_config.json, with the parameters for finetuning. Are those available anywhere? Was anyone able to finetune?

Taikakim commented 2 months ago

It's available on the Hugging Face page of Stable Audio Open: https://huggingface.co/stabilityai/stable-audio-open-1.0/tree/main

And yes, it finetunes just fine :)

cvillela commented 2 months ago

@Taikakim thank you for the reply!