Alan Akbik
Alan Akbik
Hello @geheim01 a few thoughts: - distilbert-base-uncased is for English only, so you need to use a transformer that can handle German text - if the learning rate is too...
Yes, the weird sentence printout points to the problem. Try: ` columns = {1: 'text', 2: 'ner'}` instead of ` columns = {0:'id', 1: 'form', 2: 'ner'}` and see if...
Strange, @dobbersc can you take a look?
Hello @grinay it's a bit difficult to answer, but it sounds like overfitting. Perhaps the `CharacterEmbeddings` learn very specific patterns that only exist in your training data. You could try:...
Great! You should also try decreasing the mini-batch size - if I read your output correctly you have only 20 training examples and a mini-batch size of 32 set. This...
Ah greetings back - did not realize you were at HU ;)
@lukasgarbas can you check and help him?
Hello @divyanshugit - I believe Hindi embeddings already work. Try: ```python from flair.data import Sentence from flair.embeddings import WordEmbeddings # make Hindi sentence sentence = Sentence("मैं बर्लिन में रहता हूँ")...
Hello @stefanobranco I am not seeing this behavior with your snippet. I get the following printout: ~~~ Sentence: "Vgl. Rundschreiben RAB 1/2010 Rz 8." → ["RAB"/PARTA, "1/2010"/YEAR, "Rz"/MISC, "8."/MISC] ~~~...
@stefanobranco just merged a PR that should make span detection more robust and hopefully cover your case (DOCSTART as first sentence).