qdrant / qdrant/fastembed

Does converting model to onnx format cause performance degradation of model?

Open
#554 5 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
3.2k
Forks
248
Avg merge
4d 8h
Merged PRs (30d)
4

Description

I just converted the unsupported model is Alibaba-NLP/gte-multilingual-base to onnx format to use fastembed via TextEmbedding.add_custom_model(), the result is that it works with this library. However, the problem is that I noticed that the performance of the embedding model dropped significantly when comparing the hit@1 score with the original HuggingFace model on my personal dataset, I also tried using the pre-converted model of onnx-community and the result is similar.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reproducing the reported comparison using TextEmbedding.add_custom_model() with Alibaba-NLP/gte-multilingual-base and the onnx-community model, then compare their hit@1 results with the original Hugging Face model on an equivalent dataset. Done means establishing whether conversion or model integration accounts for the degradation and documenting the evidence.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.