English NER model used for Latvian text transformations
After testing issue #3 (closed) , text transformation was seemingly working for Latvian texts. However, after using it for a while, I noticed it often replaced wrong entities. Upon further investigation I saw logs that show personal server uses some English NER model, not the new multilingual one that David mentioned.
comprise-tt logs:
2021-06-07 10:52:43,391 --------------------------------------------------------------------------------
2021-06-07 10:52:43,391 The model key 'ner' now maps to 'https://huggingface.co/flair/ner-english' on the HuggingFace ModelHub
2021-06-07 10:52:43,391 - The most current version of the model is automatically downloaded from there.
2021-06-07 10:52:43,391 - (you can alternatively manually download the original model at https://nlp.informatik.hu-berlin.de/resources/models/ner/en-ner-conll03-v0.4.pt)
2021-06-07 10:52:43,391 --------------------------------------------------------------------------------
2021-06-07 10:52:48,880 loading file /root/.flair/models/ner-english/4f4cdab26f24cb98b732b389e6cebc646c36f54cfd6e0b7d3b90b25656e4262f.8baa8ae8795f4df80b28e7f7b61d788ecbb057d1dc85aacb316f1bd02837a4a4
172.17.0.1 - - [07/Jun/2021 10:52:51] "POST /transform?%7B%22r%22:%20%22FULL%22%7D HTTP/1.1" 200 -
Edited by Dāvis Raģels