tensorflow / tensorflow/recommenders
Benchmark LightFM vs TFRS
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 2k
- Forks
- 300
- PR merge metrics
- No merged PRs in 30d
Description
Hello,
As we try both on our dataset and systematically find that LightFM outperform TFRS (for user logs and with adding metadata as well) , I am wondering if a benchmark between LightFM and TFRS have already be done and available ?
Precisely we are using the retrieval module from TFRS (as we do not have rating in our dataset but are working with explicit feedback which make sense for LightFM). with LightFM we use WARP loss.
We spent time to try optimize parameter within TFRS (including negative sampling)
The difference in very sharp in advantage of LightFM for metric like prec@n.
Does somebody make the same observation ?
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
No files or tests are named. Start with the TFRS retrieval module and compare it with LightFM using WARP loss on the same user logs, metadata, and explicit-feedback dataset, including the negative-sampling settings. Done means a reproducible benchmark reports precision@n and explains the observed difference.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, tensorflow
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 22/100