NVIDIA-Merlin / NVIDIA-Merlin/Merlin

[QST] Difference Similarity Metrics between Two tower training and QueryFaiss

Open
#889 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

question
Dominant language
Python
Stars
907
Forks
129
PR merge metrics
No merged PRs in 30d

Description

❓ Questions & Help

Details

I try to follow the example of Two Tower model training model and ensemble pipeline. However, The top-k prediction lists from two tower training and top-k search from QueryFaiss are significantly different. Which I believe due to the use of different similarity functions during training (Dot Product) and QueryFaiss (L2 Distance).

I would appreciate any advice you may have regarding this issue.

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 with the Two Tower model training example and the ensemble pipeline notebook linked in the issue. Compare the retrieval and similarity configurations used to produce each top-k list, then reproduce the discrepancy with the reported workflow. Done means the cause is documented and the two retrieval paths are either aligned or their expected difference is clearly explained.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning, search
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.