No tasks available after few accept or reject in ner.teach

[tiru1930 (Tiru)]

Hi

I am using prodigy 1.7, When I do ner.teach on my data with below command

prodigy ner.teach FR_test_3 fr_core_news_sm Fr_data.txt --patterns output.jsonl

FR_data.txt has email body per line (total 1000 emails)
output.jsonl has patterns like below

{“pattern”: [{“lower”: “suite”}, {“lower”: “\u00e0”}], “label”: “follow_up”}
{“pattern”: [{“lower”: “documents”}, {“lower”: “demand\u00e9s”}], “label”: “attach_other”}

After starting a web server with the above command ( prodigy ner.teach FR_test_3 fr_core_news_sm Fr_data.txt --patterns output.jsonl) I am seeing below problems

1. After a few iterations, I can see No tasks available on UI? Why this is happing, am I doing anything wrong
2. each Iteration I can see only on Entity, Though mail body has more entities, Is this is the expected behaviour?

Regards
Tiru

[ines (Ines Montani)]

@tiru1930

After a few iterations, I can see No tasks available on UI? Why this is happing, am I doing anything wrong

When using ner.teach, keep in mind that you're not just annotating every example. Prodigy will actively look for the most relevant examples to annotate, meaning it will skip some examples in favour of others. See this thread for more details:

"No tasks available" even though there's plenty of samples left

Thanks for the report! Just tested it with your example data and came across the same behaviour. There weren’t any errors either, so I think what might be happening here is the following: textcat.teach scores the stream and tries to only show you the most relevant tasks. Since there are only 100 examples, the “most relevant” selection seems to be only about 10-20%, which means the stream is exhausted after 10-20 examples. This is probably unideal, and we’ll think about the best way to handle t…

@tiru1930

each Iteration I can see only on Entity, Though mail body has more entities, Is this is the expected behaviour?

Yes, in ner.teach, you're collecting binary feedback on specific annotation suggestions. So you're accepting or rejecting the currently highlighted entity span.