-
Notifications
You must be signed in to change notification settings - Fork 21
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.
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
Fix things, prepare for v0.2 #85
Conversation
return metadata | ||
path = config.path / f"metadata.yaml" | ||
logger.warning(f"Loading metadata from {path}") | ||
return CheckpointMetadata.from_dict(yaml.safe_load(path.open("r"))) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@charlesGE can you confirm that we don't need this anymore please?
@pytest.mark.depends(on=["test_model_sf"]) | ||
def test_model_sdp2(): | ||
# Sequence-data-parallel | ||
run_test_script( |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
+1
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM, but I'd like to defer to @charlesGE who should try to work with ghcr.io/servicenow/fast-llm:sha-85dfeea
and see if our experiments can be continued with it.
✨ Description
check_parallel_match
debug_tensor_parallel
for sequence-tensor-parallelpresents
wrongly set for the last micro-batch with sequence-data-parallel, causing a crash ([bug] Crash with sequence-data-parallel #57)tensor_and_sequence_data
distributed dim so we can reconstruct micro-sequences properly (needed for test below)🔍 Type of change
Select all that apply: