google-deepmind / google-deepmind/scalable_agent
How to understand the multi-learners in IMPALA
Open
- Dominant language
- Python
- Stars
- 1k
- Forks
- 160
- PR merge metrics
- No merged PRs in 30d
Description
*Multiple Synchronous Learners* are mentioned in Figure 1 of IMPALA's paper, but I find no clear multi-learners features in the actual codes. May I ask how the codes implements the structure of *Worker Learner - Master Learner* as mentioned in Figure 1?
Contributor guide
Assessment
This issue has not been assessed yet.