truth should be a list of list of strings after transform which are non-empty

[its-ogawa]

Use the latest huggingsound.

#!pip list | grep huggingsound
huggingsound 0.1.4

When I run evaluate on a finetuned model, I get a ValueError.

/usr/local/lib/python3.7/dist-packages/jiwer/measures.py in _preprocess(truth, hypothesis, truth_transform, hypothesis_transform) 332 if not _is_list_of_list_of_strings(transformed_truth, require_non_empty_lists=True): 333 raise ValueError( --> 334 "truth should be a list of lists of strings after transform which are non-empty" 335 ) 336 if not _is_list_of_list_of_strings(

ValueError: truth should be a list of list of strings after transform which are non-empty

However, I do not think it is a bug in evaluate, because it terminates successfully for the trained model obtained as is, as shown below. https://github.com/jonatasgrosman/huggingsound#evaluation

I think something is going on during finetune, what does this message point me to do? Any advice would be appreciated.

[Shaobo-Z]

I am having the same issue. It looks like the data format passed to the function is incorrect. Did not find a solution for that.