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

Keep max_seqlen and cu_seqlens_argmin for later micro-batches when PP>1 #8334

Merged
merged 2 commits into from
Feb 6, 2024
Merged

Keep max_seqlen and cu_seqlens_argmin for later micro-batches when PP>1 #8334

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
9201c51
Keep max_seqlen and cu_seqlens_argmin for later micro-batches when PP>1
erhoo82 Feb 2, 2024
f618405
Merge branch 'r1.23.0' into sft_no_sync
ericharper Feb 6, 2024
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/models/language_modeling/megatron_gpt_model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -943,8 +943,8 @@ def fwd_output_and_loss_func(dataloader_iter, model, checkpoint_activations_all_

# Transfer needed data to GPU
required_keys = set()
max_seqlen = batch.pop('max_seqlen').squeeze() if 'max_seqlen' in batch else None
cu_seqlens_argmin = batch.pop('cu_seqlens_argmin') if 'cu_seqlens_argmin' in batch else None
max_seqlen = batch['max_seqlen'].squeeze() if 'max_seqlen' in batch else None
cu_seqlens_argmin = batch['cu_seqlens_argmin'] if 'cu_seqlens_argmin' in batch else None
if parallel_state.get_pipeline_model_parallel_world_size() == 1:
required_keys.update(batch.keys())
else:
Expand Down
Loading