About training setups

[jee019]

Hi, thank you for sharing this great work.

I have a couple of questions regarding the training setup:

1. Could you provide more details on the training configuration? Specifically, were all the released checkpoints trained strictly using the config files currently in the repository, or were there any modifications?
2. From the example provided, it seems like 8 GPUs were used for training. Assuming a batch size of 8 and using the HuBERT base model to train X-Codec, could you provide a rough estimate of the total training time required?

Thank you in advance for your help!