search

wenet-e2e / wenet

Production First and Production Ready End-to-End Speech Recognition Toolkit
https://wenet-e2e.github.io/wenet/
Apache License 2.0
4.16k stars 1.08k forks source link

whisper script does not work #2626

Open 200987299 opened 2 months ago

200987299 commented 2 months ago

loss_dict = executor.cv(model, cv_data_loader, configs) File "/datasets_hdd/dxu/wenet/wenet/utils/executor.py", line 161, in cv loss_dict["acc"] = sum(total_acc) / len(total_acc) ZeroDivisionError: division by zero return f(*args, **kwargs) File "/datasets_hdd/dxu/wenet/examples/common_voice_yue_t2/whisper/wenet/bin/train.py", line 162, in main loss_dict = executor.cv(model, cv_data_loader, configs) File "/datasets_hdd/dxu/wenet/wenet/utils/executor.py", line 161, in cv loss_dict["acc"] = sum(total_acc) / len(total_acc) ZeroDivisionError: division by zero

That is the error I got, I tried two datasets but they all gives me the same error. BTW, the "input_dim" parameter is missing from the default config file.

Please let me know what I did wrong or maybe there is a bug somewhere?

bests

turinaf commented 1 month ago

I'm also facing this error

turinaf commented 1 month ago

@200987299 were you able to solve?