Skip to content

skip previous trained batches#42

Open
aws-zhenguo wants to merge 1 commit intomain_11-07-24_gpu_baselinefrom
replay_gpu
Open

skip previous trained batches#42
aws-zhenguo wants to merge 1 commit intomain_11-07-24_gpu_baselinefrom
replay_gpu

Conversation

@aws-zhenguo
Copy link
Copy Markdown
Collaborator

No description provided.

@aws-zhenguo aws-zhenguo requested a review from HahTK December 23, 2024 19:02
Copy link
Copy Markdown

@satishpasumarthi satishpasumarthi left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Left a comment

Comment thread axlearn/common/trainer.py

# skip previous batches
for i in range(self.step):
self.vlog(3, f"skipping batch for step {i}")
Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do we need to add any unit tests for this change?

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No unit tests planed for now since these are not merging to upstream.

@aws-zhenguo aws-zhenguo force-pushed the replay_gpu branch 2 times, most recently from 2fd02ba to 1759e72 Compare December 23, 2024 20:43
Copy link
Copy Markdown

@satishpasumarthi satishpasumarthi left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM!

@HahTK
Copy link
Copy Markdown
Collaborator

HahTK commented Dec 27, 2024

LGTM.
Needs a test that we can run on both GPU and TRN

@aws-zhenguo
Copy link
Copy Markdown
Collaborator Author

LGTM. Needs a test that we can run on both GPU and TRN

tested in both GPU and TRN. should we merge it?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants