Hi Yuan, thanks for the awesome contributions here. I was wondering if you have the ESC-50 model (results reported in the original paper) released anywhere, as I have not been able to find it through this repo or on HF. I see the recipe for re-training on ESC-50 in the README, but wondering if there is a checkpoint of this fine-tuned version (with a 50-class head) that I'm missing somewhere.
Thank you!