You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Actually, I want to extract the features from the AV-HuBERT, so I decided to use the "extract_finetune".
But, I just would like to understand the difference between the functions, "extract_finetune" and "extract_features".
In the Colab demo, it appears that "extract_finetune" is used to obtain features regardless of whether the model and checkpoint are pre-trained or fine-tuned. Can I use the "extract_finetune" function for both pre-trained and fine-tuned models and checkpoints?
Additionally, I would like to extract the discrete unit feature from AV-HuBERT, not the continuous features obtained via "extract_finetune". Could you provide guidance on extracting features similar to the unit feature described in this paper?: https://arxiv.org/pdf/2204.02967
Hi, Thank you for your dedicated work.
Actually, I want to extract the features from the AV-HuBERT, so I decided to use the "extract_finetune".
But, I just would like to understand the difference between the functions, "extract_finetune" and "extract_features".
In the Colab demo, it appears that "extract_finetune" is used to obtain features regardless of whether the model and checkpoint are pre-trained or fine-tuned. Can I use the "extract_finetune" function for both pre-trained and fine-tuned models and checkpoints?
Additionally, I would like to extract the discrete unit feature from AV-HuBERT, not the continuous features obtained via "extract_finetune". Could you provide guidance on extracting features similar to the unit feature described in this paper?: https://arxiv.org/pdf/2204.02967
Thank you for your assistance.