KeyError: 'file' in prediction

[Efr0nd]

Thank you for providing the code! I encountered this issue while running the command 'prediction.sh output/tmp':

_File "run_emotion.py", line 599, in main() File "run_emotion.py", line 581, in main f.write(val_dataset[i]['file'].split("/")[-1] + " " + str(len(val_dataset[i]['inputvalues'])/16000) + " ") KeyError: 'file'

And I checked the keys of the val_dataset and found only 'input_value'. Why did this happen? After mapping, val_dataset[i] still has the key 'file'. It has truly been troubling me for quite a while.

I would greatly appreciate it if you could answer my question!

[Efr0nd]

And for this line of code:

predictions, labels, metrics = trainer.predict(val_dataset, metric_key_prefix="predict")

I found many 'Nan' in predictions, and the labels is None.

[TideDancer]

Sorry for the late reply. Have you solved the issue? The csv file has the field 'file' in it. If the 'file' key is not found, could you set a breakpoint after obtaining the val_dataset, https://github.com/TideDancer/interspeech21_emotion/blob/c36dcc0d2bd9a22602c081a5ab064ab5e9d4f019/run_emotion.py#L448, and see if the keys are correct ?

[Efr0nd]

Sorry for the late reply. Have you solved the issue? The csv file has the field 'file' in it. If the 'file' key is not found, could you set a breakpoint after obtaining the val_dataset,

https://github.com/TideDancer/interspeech21_emotion/blob/c36dcc0d2bd9a22602c081a5ab064ab5e9d4f019/run_emotion.py#L448

, and see if the keys are correct ?

Thanks for your reply! I will try that.

[wlaamaa]

Sorry for the late reply. Have you solved the issue? The csv file has the field 'file' in it. If the 'file' key is not found, could you set a breakpoint after obtaining the val_dataset, https://github.com/TideDancer/interspeech21_emotion/blob/c36dcc0d2bd9a22602c081a5ab064ab5e9d4f019/run_emotion.py#L448

, and see if the keys are correct ?

Thanks for your reply! I will try that.

Hello, have you solved this problem