Scoring method
#16
by
Iantsa
- opened
Hello,
i have a question, in the inference API , which scoring method is used? cosine_similarity?
my problem was, with large sentence, inference API and my local model didnt give the same score,
but i just saw the previous discussion by the community,
this is because of tokenizer max_length, i put it to 128, and it gave the same score, so i supose that the scoring method is the same: cosine_similarity