We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
It's becoming the norm to have prompt prefixes for text embedding models. I think we should add this to the hf-embedder.
<component id="snow" type="hugging-face-embedder"> <transformer-model url="https://huggingface.co/Snowflake/snowflake-arctic-embed-l/resolve/main/onnx/model_int8.onnx"/> <tokenizer-model url="https://huggingface.co/Snowflake/snowflake-arctic-embed-l/raw/main/tokenizer.json"/> <normalize>true</normalize> <pooling-strategy>cls</pooling-strategy> <instruction-prompt> <query>Represent this sentence for searching relevant passages:</query> <document>passage:</document> </instruction-prompt> </component>
The embedder would then prepend the input with these instructions depending on the context (query or indexing)
The text was updated successfully, but these errors were encountered:
Alternatives
<instruction-prompts> <query>Represent this sentence for searching relevant passages:</query> <document>passage:</document> </instruction-prompt>
<prefixes> <query>Represent this sentence for searching relevant passages:</query> <document>passage:</document> </prefixes>
<prepend query="Represent this sentence for searching relevant passages:" document="passage:"/>
<prepend> <query>Represent this sentence for searching relevant passages:</query> <document>passage:</document> </prepend>
Since normalize is a verb I think that prepend is a good alternative, so I'm voting for alternative 4.
normalize
Sorry, something went wrong.
jobergum
No branches or pull requests
It's becoming the norm to have prompt prefixes for text embedding models. I think we should add this to the hf-embedder.
The embedder would then prepend the input with these instructions depending on the context (query or indexing)
The text was updated successfully, but these errors were encountered: