Named entity recognition (NER) model evaluation

In this video we are going to talk about different ways how one can evaluate an NER (named entity recognition) model.


Code from the video:
https://github.com/jankrepl/mildlyoverfitted/tree/master/github_adventures/ner_evaluation
https://github.com/chakki-works/seqeval

00:00 Intro
00:31 Mispredictions
02:31 IOB2 notation
04:03 Evaluation approaches
07:38 [code] HF evaluate + seqeval
14:36 [code] Enitity-level from scratch intro
17:38 [code] Check sequence valid
23:23 [code] Extract entity types
24:42 [code] Extract entities
32:33 [code] Generate report
41:06 Outro

If you have any video suggestions or you just wanna chat feel free to join the discord server: https://discord.gg/a8Va9tZsG5

Twitter: https://twitter.com/moverfitted

Тэги:

#machine_learning #natural_language_processing #nlp #named_entity_recognition #seqeval #hugging_face #evaluation #metric #classification #precision #f1score #recall #classification_report #scikit-learn #sklearn #python #tutorial #pytest #tmux #vim #hands_on #screencast #live_coding #artificial_inteligence #spacy #ner

Скачать видео