[BUG] "Spike no more" feature was removed
Open
bug
community-request
waiting-on-maintainers
- Dominant language
- Python
- Stars
- 17.9k
- Forks
- 4.5k
- Avg merge
- 4d 6h
- Merged PRs (30d)
- 271
Description
https://github.com/NVIDIA/Megatron-LM/pull/2047
this PR removed this feature (https://github.com/NVIDIA/Megatron-LM/blob/v0.15.3/megatron/training/arguments.py#L2302)
also see "default_skip_embedding_weight_decay"
Contributor guide
Assessment
This issue has not been assessed yet.