Model Training Error

[kylethieringer]

Describe the bug When trying to train new model I am running into an error in the random under sampling function.

To Reproduce Steps to reproduce the behavior:

1. load project config
2. train single model

Screenshots

Desktop (please complete the following information):

• OS: ubuntu 22.04
• Python Version 3.6
• Are you using anaconda? yes

Additional context Im not sure if this is important but I uploaded tracks from sleap h5 files and labels from deepethogram. I have 2 animals with 13pts each. I am trying to classify a sparse behavior <10% of frames.

[sronilsson]

Thank you @kylethieringer! I inserted the bout sampler train/test option recently, haven't had any feedback yet, so might be a few bugs to clean up, so this is helpful.

The error came from the bout sampling function returning arrays, and the downstream random undersampler accepting dataframes. If both bout sampling and the random undersampler was used, you'd bump into that error.

I fixed it in version 1.51.4 - could you update with pip install simba-uw-tf-dev --upgrade and see if it works on your end?

[kylethieringer]

Thanks so much for your speedy response and fix! No issues anymore!