hello, author, i have little knowledge about nlp, and i failed to run the code ./bert_emb, i guess may be this line is wrong, for output only have 2 dim.(you could also check other line)
|
hidden_states = outputs[2] |
at the same time, i want to ask you about the cosine similarity of the txt embedding, the txt embedding trained in your code is 0.83, is this a good embedding (i know little about nlp)?
looking forward to your apply, thank you
hello, author, i have little knowledge about nlp, and i failed to run the code ./bert_emb, i guess may be this line is wrong, for output only have 2 dim.(you could also check other line)
StackGAN-BERT/input/src/bert_emb.py
Line 53 in bfb1596
at the same time, i want to ask you about the cosine similarity of the txt embedding, the txt embedding trained in your code is 0.83, is this a good embedding (i know little about nlp)?
looking forward to your apply, thank you