F1 score of 0 for character level sequence labelling task

Training data looks like:

and training results look like:

Also worth adding that whenever I add a new period at the end of each character sequence, it never begins training and stops at Epoch 1/15 with no progress bar. I’ll be happy to share training data with you if that will help reproduce the issue.

[tc64]

I think you cannot just use any tag label schema and expect evaluation function to work correctly, without editing anago.metrics.get_entities - it expects BILO.

[Hironsan]

Not use BILO. Use IOB2.

Because anago ver0.0.5 assumes that training data is labeled by IOB2.

I will fix the problem in ver 1.0.0.