Hugging Face's logo

BAAI
/
bge-m3

Sentence Similarity
sentence-transformers
PyTorch
ONNX
xlm-roberta
feature-extraction
Eval Results
text-embeddings-inference
Model card Files
xet
 Community
144

Do I need to add the prefix "query: " and "passage: " to input texts?

#35
by zaobao - opened
Discussion
zaobao

As the query prefix of 'bge-large-zh-v1.5' is "为这个句子生成表示以用于检索相关文章：", should I add the same prefix to queries with this model ?
Shitao
 Beijing Academy of Artificial Intelligence org

No. bge-m3 doesn't need prefix instruction.
zaobao
•
edited Mar 31, 2024

So the model supports both query-to-query and query-to-answer-retrieval without any prefixes or instructions. Did I get it right ?
Shitao
 Beijing Academy of Artificial Intelligence org

So the model supports both query-to-query and query-to-answer-retrieval without any prefixes or instructions. Did I get it right ?

Yes, it supports both query-to-query and query-to-answer-retrieval.
TuyoSera

So the model supports both query-to-query and query-to-answer-retrieval without any prefixes or instructions. Did I get it right ?

Yes, it supports both query-to-query and query-to-answer-retrieval.

Does it need normalizing embeddings?
rajydv

So the model supports both query-to-query and query-to-answer-retrieval without any prefixes or instructions. Did I get it right ?

Yes, it supports both query-to-query and query-to-answer-retrieval.

Does it need normalizing embeddings?

By default, bge-m3 embeddings are normalized (ref)

Sign up or log in to comment