Hello!
I've been trying to train a model using my own data, but I ran into an error when running the trainscript file:
Epoch #1
yee_bach_passion#2
[Instrument(program=0, is_drum=False, name="")]
yee_bach_passion#41
[Instrument(program=0, is_drum=False, name="")]
...
yee_bach_passion#1
[Instrument(program=0, is_drum=False, name="")]
yee_bach_passion#21
[Instrument(program=0, is_drum=False, name="")]
Traceback (most recent call last):
File "midi2params/trainscript.py", line 409, in <module>
main()
File "midi2params/trainscript.py", line 155, in main
batch = trim_and_preprocess(batch, config)
File "/home/poohcom1/midi2params_env/midi2params/midi2params/train_utils.py", line 93, in trim_and_preprocess
subclip_start = np.random.randint(50, final_idx - 50 - len_subclip)
File "mtrand.pyx", line 748, in numpy.random.mtrand.RandomState.randint
File "_bounded_integers.pyx", line 1247, in numpy.random._bounded_integers._rand_int64
ValueError: low >= high
This always happens during the first epoch after the program printed a few file.
I'm still not familiar with the code base so I'm not sure how the upper limit is being calculated, and whether the issue has something to do with my dataset, so any pointers would be appreciated!
Cheers