hi ,i want to use flair embedding and use myself model instead of the API function ,how should i do , thanks.anyone can give me a advice , thanks a lot.

[cpmss521]

A clear and concise description of what you want to know.

[stefan-it]

Hi @cpmss521,

does it help when you use the embeddings for each token of an input sequence?

E.g. you have the sentence "Berlin and Munich are nice cities.". Then you can use the Sentence class + its internal tokenizer:

from flair.data import Sentence
from flair.embeddings import FlairEmbeddings

input_sequence = "Berlin and Munich are nice cities."

sentence = Sentence(text=input_sequence, use_tokenizer=True)

# Forward language model
forward_embeddings = FlairEmbeddings('news-forward')

# Backward language model
backward_embeddings = FlairEmbeddings('news-backward')

# Get embeddings for each token in sentence (both forward and backward language model)
forward_embeddings.embed(sentence)
backward_embeddings.embed(sentence)

for token in sentence:
    # Dimension of forward language model is 2048
    # Same as for the backward language model
    # So final embedding vector for each token is 4096
    print(token.get_embedding())

I hope this helps you 😃

[cpmss521]

thanks a lot ! i get it by document,but i have some confusion about :
#502 (comment)

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.