RNNT模型在executer执行cv操作的时候提示KeyError：“th_accuracy”

[DaobinZhu]

训练完毕后执行cv操作的时候报错，但是看代码感觉没有问题
2024-02-02 11:31:51,367 DEBUG TRAIN Batch 0/7500 loss 129.710220 loss_att 106.460495 loss_ctc 154.101700 loss_rnnt 131.107971 lr 0.00030000 grad_norm 27.183880 rank 1
2024-02-02 11:31:51,367 DEBUG TRAIN Batch 0/7500 loss 149.184647 loss_att 124.107864 loss_ctc 157.846741 loss_rnnt 153.045059 lr 0.00030000 grad_norm 27.183880 rank 2
2024-02-02 11:31:51,367 DEBUG TRAIN Batch 0/7500 loss 140.741501 loss_att 112.667061 loss_ctc 158.834442 loss_rnnt 143.944000 lr 0.00030000 grad_norm 27.183880 rank 3
2024-02-02 11:31:51,368 DEBUG TRAIN Batch 0/7500 loss 129.564987 loss_att 106.934181 loss_ctc 157.959274 loss_rnnt 130.305237 lr 0.00030000 grad_norm 27.183880 rank 0
Traceback (most recent call last):
File "wenet/bin/train.py", line 175, in
main()
File "/home/lsj/.conda/envs/rnnt/lib/python3.8/site-packages/torch/distributed/elastic/multiprocessing/errors/init.py", line 346, in wrapper
return f(*args, **kwargs)
File "wenet/bin/train.py", line 147, in main
loss_dict = executor.cv(model, cv_data_loader, configs)
File "/home/lsj/zdb/biye/wenet/wenet/utils/executor.py", line 123, in cv
) if _dict['th_accuracy'] is not None else 0.0)
KeyError: 'th_accuracy'
Traceback (most recent call last):
File "wenet/bin/train.py", line 175, in
main()
File "/home/lsj/.conda/envs/rnnt/lib/python3.8/site-packages/torch/distributed/elastic/multiprocessing/errors/init.py", line 346, in wrapper
return f(*args, **kwargs)
File "wenet/bin/train.py", line 147, in main
loss_dict = executor.cv(model, cv_data_loader, configs)
File "/home/lsj/zdb/biye/wenet/wenet/utils/executor.py", line 123, in cv
) if _dict['th_accuracy'] is not None else 0.0)
KeyError: 'th_accuracy'

[xingchensong]

可以在这里的dict中加一下th_accuracy
https://github.com/wenet-e2e/wenet/blob/main/wenet/transducer/transducer.py#L144-L148

修复后可以提个PR

[DaobinZhu]

可以在这里的dict中加一下th_accuracy https://github.com/wenet-e2e/wenet/blob/main/wenet/transducer/transducer.py#L144-L148

修复后可以提个PR

好的，感谢大佬

[DaobinZhu]

犯了个低级错误，居然在看asr_model.py，已经提交PR了 #2337