Menglong Xu

Aboult the warmup_n_steps and accum_grad_n_steps usd in Aishell

1

In the config transformer.yaml, "n_epochs=30, batch_size=32, warmup_n_steps=25000, accum_grad_n_steps=8". Under the above config, I got a learning rate curve which went straight up until the end of the training. And the...

Great work! I'd appreciate it if you could tell me the wer/cer of this project on the various mainstream corpuses.

Any plans for adding support for PyTorch or onnx?

1

Hi there, Really great to see your work! Just wondering do you have any plans to support PyTorch model or onnx model?

implementation of fsmn

1

https://github.com/wenet-e2e/wekws/blob/1a8ee651620f49a8a7779be4d30668157fc23350/wekws/model/fsmn.py#L225 为什么只在一侧对input进行pad