I wonder why only 'Nor' is displayed in dls.show_batch(sharvey=True).

[DongHyukHwang]

I'm running https://colab.research.google.com/github/timeseriesAI/tsai/blob/master/tutorial_nbs/01a_MultiClass_MultiLabel_TSClassification.ipynb line by line.

I am using the latest version of tsai 0.3.0.
In the middle of following each line, you can see that all of my result screens are displayed as 'Nor' in the

Visualize data
dls.show_batch(sharey=True) part.

You can see that the 'ROT' value is also mixed in the web page. I wonder why this issue occurred.

[oguiza]

Hi @DongHyukHwang,
Thanks for creating this issue. It's actually kind of a bug in the code.
I'll explain. In order to speed up training, some array-like structures (like zarr) require samples to be sorted. Sorting is implemented within each batch. But there's a downside. If the samples in the dataset are sorted by class, only the first class/es will be displayed. I've flagged this as a bug and will see how it can be fixed.

[oguiza]

Hi @DongHyukHwang,
I've already fixed the code in GitHub. It'd be good if you could test that show_batch works well now.
You'll need to pip install using:

pip install git+https://github.com/timeseriesAI/tsai.git

[DongHyukHwang]

[DongHyukHwang]

Could you change the encoding to cp949?

[oguiza]

This seems to be a new issue unrelated to the original one. If it still persists, please create a new issue.