tensorflow / tensorflow/recommenders

The floor value for clipping in SamplingProbablityCorrection could be too high for larger datasets

Open
#588 0 comments 4 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
2k
Forks
300
PR merge metrics
No merged PRs in 30d

Description

Hi,

The current value of 1e-6 is set here. The rationale seems to be to avoid zeros but the value set could be too high for larger datasets and result in lower metrics on a private dataset. This stops us from updating the package to versions >0.6.0. What would be the best way to address it?

One way could be to set a lower value as suggested here. Also, there seems to be a typo in the naming of the class that's already addressed in another PR that hasn't been merged yet

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

Locate the SamplingProbablityCorrection class and inspect the 1e-6 clipping value described in the issue. Read PR 567 for the proposed lower value and PR 300 for the class-name typo, then determine the intended correction and how it should be validated. Done means the clipping behavior is agreed and covered without harming larger-dataset metrics.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, tensorflow
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
38/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.