Batch inference

[anhnh2002]

How can i preprocess inputs (eg: padding) for batch inference?

[jishengpeng]

How can i preprocess inputs (eg: padding) for batch inference?

The current code framework does not support batch reasoning because the training process is randomly selected for one second. You can split the inference data set or try to multi-process to achieve the goal of multiple batches in the inference

[anhnh2002]

How can i preprocess inputs (eg: padding) for batch inference?

The current code framework does not support batch reasoning because the training process is randomly selected for one second. You can split the inference data set or try to multi-process to achieve the goal of multiple batches in the inference

Can I adjust the code for batch inference (e.g., padding the input, then truncating codes using a padding mask)? If so, how can I do it?

[jishengpeng]

How can i preprocess inputs (eg: padding) for batch inference?

The current code framework does not support batch reasoning because the training process is randomly selected for one second. You can split the inference data set or try to multi-process to achieve the goal of multiple batches in the inference

Can I adjust the code for batch inference (e.g., padding the input, then truncating codes using a padding mask)? If so, how can I do it?

You can add padding when inference, but you must record the corresponding mask position, and make sure that each neural network module during inference only deals with the non-zero part of the audio