openspeech

openspeech copied to clipboard

# ❓ Questions & Help Thanks for your great works on ASR. I am wondering if we have training result data for provided model? I think it is important for...

Peach-He

Potential Issue with EncoderDecoder Model with CrossEntropy Loss

2

Hi, I see that the LSTM Attention Decoder takes the log_softmax of the step outputs [inside the model](https://github.com/openspeech-team/openspeech/blob/main/openspeech/decoders/lstm_attention_decoder.py#L155). However, cross entropy loss uses [nn.CrossEntropyLoss](https://github.com/openspeech-team/openspeech/blob/main/openspeech/criterion/cross_entropy/cross_entropy.py#L108), which takes another log_softmax inside. Shouldn't...

gkeskin07

### # ❓ Questions & Help Seq2Seq 모델에서 Teacher forcing 사용시 validation loss가 지나치게 높고, epoch당 validation loss가 증가함 ### environment Device: RTX 3090 * 4 OS: Ubuntu 20.24 CUDA...

Klassikcat

prediction(inference)

1

# ❓ Questions & Help ## Details hi, do you have prediction script?

ggyuu1

Isn't Conv1D with kernel=1 the same as Linear layer but slower?

1

# ❓ Questions & Help I was wandering whether it makes sense from an efficiency point of view to change the Conv1D layer with kernel 1 from the conformer encoder...

OleguerCanal

Resuming from a checkpoint that ended before the epoch ended

2

# ❓ Questions & Help epoch 한번이 끝나기 전 런타임이 끊겼는데요. 혹시 이런 상황에선 checkpoint를 불러오는 것이 불가능할까요 아래와 같은 메세지가 떴습니다! ## Details ``` UserWarning: You're resuming from a...

rkskekzzz

AssertionError: choose a window size 320 that is [2, 0]

1

# ❓ Questions & Help AssertionError: choose a window size 320 that is [2, 0], please tell me how to solve it. My command is python ./openspeech_cli/hydra_train.py dataset=aishell dataset.dataset_path=/anlab/stuhome/zngz28/access/aishell_dataset/data_aishell dataset.dataset_download=False...

Yonghao-Tan

Loss becomes NAN after a while

3

## Environment info - Platform: Ubuntu20.04 - Python version: 3.8.10 - PyTorch version (GPU?): 1.9.0+cu111 (pytorch lightning 1.5.8) - Using GPU in script?: 4x A100 ## Information Model I am...

OleguerCanal

Why are the error rate accumulated?

5

# ❓ Questions & Help  ## Details Each call of the error rate accumulates the distance and length. Why is that?Is it to have a running average kind of...

OleguerCanal

Am I training properly?

21

I am not familiar with ASR tasks, so I'd be so glad if anyone answer my question: I am training ContextNet which is basically rnn-t type model as in the...

resurgo97