"mtrand.pyx", line 1120, in mtrand.RandomState.choice ValueError: a must be greater than 0

[vidigen]

First of all thx a lot for posting this very interesting work.

"mtrand.pyx", line 1120, in mtrand.RandomState.choice
ValueError: a must be greater than 0

In running the maml.py I incurred in the error above. I believe is a numpy error... How do you get around it?

Thanks a lot,

[vidigen]

to be more precise it gets stuck in line 80 of core.py code:
query = df[(df['class_id'] == k) & (~df['id'].isin(support_k[k]['id']))].sample(self.q)

[Shandilya21]

Hi, @oscarknagg thanks for posting this work.
Hi @vidigen I have similar issues, in, it raises a different error which is something like

File "../few_shot/train.py", line 108, in fit

for batch_index, batch in enumerate(dataloader):

File "/home/shandilya/nlp_project/lib/python3.5/site-packages/torch/utils/data/dataloader.py", line 193, in iter
return _DataLoaderIter(self)
File "/home/shandilya/nlp_project/lib/python3.5/site-packages/torch/utils/data/dataloader.py", line 493, in init
self._put_indices()
File "/home/shandilya/nlp_project/lib/python3.5/site-packages/torch/utils/data/dataloader.py", line 591, in _put_indices
indices = next(self.sample_iter, None)
File "../few_shot/core.py", line 80, in iter
query = df[(df['class_id'] == k) & (~df['id'].isin(support_k[k]['id']))].sample(self.q)
File "/home/shandilya/nlp_project/lib/python3.5/site-packages/pandas/core/generic.py", line 2144, in sample
locs = rs.choice(axis_length, size=n, replace=replace, p=weights)
File "mtrand.pyx", line 1165, in mtrand.RandomState.choice

ValueError: Cannot take a larger sample than population when 'replace=False'

Did you know the solution for this problem? Initially, I thought it was the problem related to the NumPy sample (with or without replacement)? But as you said, it was associated with few_shot/core.py raises an error at line 80.

Thanks for your help and I am looking for your comment.

[Sgsouham]

@oscarknagg any updates on this??

[ucalyptus]

@Shandilya21 did you solve this ?

[ceevaaa]

Which dataset you guys are trying to run this?

[AlisonYao]

Solved this problem by deleting the hidden .DS_store file in my images_evaluation folder. There could also be a .DS_store file in the images_background folder (somehow I don't have it though). It seems that whenever episode_classes has the class 0, the path is directed to .DS_store instead of the actual data folder, which caused an error.