search

YuanGongND / ssast

Code for the AAAI 2022 paper "SSAST: Self-Supervised Audio Spectrogram Transformer".
BSD 3-Clause "New" or "Revised" License
365 stars 61 forks source link

Host on Huggingface #24

Open penguinwang96825 opened 1 year ago

penguinwang96825 commented 1 year ago

Have you ever considered hosting the pre-trained SSAST model checkpoints on the Huggingface Hub?

YuanGongND commented 1 year ago

hi there,

AST is on HF, not SSAST. I do not have time to convert the SSAST codo to HF.

-Yuan

SYoy commented 4 months ago

Hi @penguinwang96825,

If you are interested in using SSAST model weights with the HuggingFace ecosystem, you may find my recent blog post helpful. It walks through the entire process of loading a model weights from this repository into the ASTModel and ASTForAudioClassification implementations of the Transformers library.

https://medium.com/itnext/how-to-use-ssast-model-weigths-in-the-huggingface-ecosystem-0f3fdc8d38da

By the way, thanks for the awesome model, Yuan.

Cheers Syoy