search

facebookresearch / audiocraft

Audiocraft is a library for audio processing and generation with deep learning. It features the state-of-the-art EnCodec audio compressor / tokenizer, along with MusicGen, a simple and controllable music generation LM with textual and melodic conditioning.
MIT License
20.59k stars 2.09k forks source link

inference with custom weights #219

Open icklerly1 opened 1 year ago

icklerly1 commented 1 year ago

Hello,

I have successfully trained a custom model. Now I have some .th files in my train folder but I am not sure how to load the models and do an inference.

Can you give me a hint how to load the model?

This is the content of my folder in total:

Bildschirm­foto 2023-08-14 um 08 36 12
ElizavetaSedova commented 1 year ago

@icklerly1 Hello, I think the instructions for exporting your model will help you. https://github.com/facebookresearch/audiocraft/blob/main/docs/MUSICGEN.md#importing--exporting-models

cvillela commented 1 year ago

@Elizahub how can one get the SIG of a pretrained model?

icklerly1 commented 1 year ago

Yeah, I am wondering exactly the same thing :) What does the "SIG_OF_LM" mean?

cvillela commented 1 year ago

The SIG is just the hash created to represent the dora experiment, such as 71bdb194. But the concept of SIG vs XP are still kind of loose on me.

adefossez commented 1 year ago

@icklerly1 in your case the SIG is ea5ff7e7