deepinsight / deepinsight/insightface
Learning rate step strategy
Open
- Dominant language
- Python
- Stars
- 29.7k
- Forks
- 6.1k
- PR merge metrics
- No merged PRs in 30d
Description
The original article says that the learning rate for the Glint360 dataset divided by 10 at 200k, 400k, 500k, 550k iterations
and finish at 600K iterations. Are there any recommendations for a strategy to reduce LR for datasets of 1 million, 10 million, 100 million identities? Thanks in advance for your reply
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.