Batch size exceeds the corpus length

[lyvnviinky]

keep getting "AssertionError: Batch size exceeds the corpus length", even when i set batch_size to 1
I'm trying to run this locally on windows with anaconda3
My dataset has 919 wav files, and i set the samplerate to 32000, but i don't think this should be an issue.
I don't know what's wrong, but i'll gladly give more info if needed

[relativeflux]

@lyvnviinky OK, that's interesting - was this dataset created with the chunk_audio.py script? And what is the sample rate of the files in the dataset? The sample_rate param to train.py is just for the audio files generated during training.

The line causing that AssertionError is in the get_dataset_filenames_split function in dataset.py, line 30 in that file. You could try commenting it out to see what's actually happening (maybe print len(files) to see what it thinks the value of that is - it should just be the number of files in the folder, 919 in this case).

Could you post how your're calling train.py?