Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Support O2 training of PEFT and SFT #7971

Merged
merged 2 commits into from
Dec 5, 2023
Merged

Support O2 training of PEFT and SFT #7971

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f1c9844
support O2
cuichenx Dec 5, 2023
4c41c6b
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] Dec 5, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions nemo/collections/nlp/parts/mixins/nlp_adapter_mixins.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -67,7 +67,7 @@ def __init__(self, *args, **kwargs):
self.use_ptuning_only = False
super().__init__(*args, **kwargs)
if hasattr(self, "enc_dec_model"):
self.model_prefix = "enc_dec_model." # for T5
self.model_prefix = "enc_dec_model.module." if self.cfg.megatron_amp_O2 else "enc_dec_model." # for T5
else:
self.model_prefix = "model.module." if self.cfg.megatron_amp_O2 else "model."

Expand Down Expand Up @@ -351,7 +351,7 @@ def sharded_state_dict(self, prefix: str = ''):
if not use_mcore_gpt or (self.use_peft and self.setup_complete):
return None
else:
return self.model.sharded_state_dict(prefix=self.model_prefix)
return super().sharded_state_dict(prefix=prefix)

def load_state_dict(self, state_dict, strict: bool = True):
if len(state_dict) == 0:
Expand Down
Loading