tensorflow / tensorflow/recommenders
Question - Anyone Trained Retrieval Distributed Successfully?
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 2k
- Forks
- 300
- PR merge metrics
- No merged PRs in 30d
Description
I am curious is anyone out here has successfully run a basic two tower retrieval model distributing using Horovod or any other method? I am seeing consistently poor results training in a distributed fashion and better with a single GPU (although not incredible). For the Horovod method I am sharding the interaction data across 8 GPUs. Has anyone seen good performance distributing this computation?
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
The issue names no files, tests, or entry points. Start by reviewing the two-tower retrieval training setup using Horovod and interaction data sharded across eight GPUs; done would mean identifying the cause of poorer distributed results or documenting a successful distributed approach.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, tensorflow
- Domain
- distributed-systems, machine-learning
- Issue type
- Bug
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100