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

Better ElasticDeviceMesh #9

Merged
merged 33 commits into from
Sep 28, 2024
Merged

Better ElasticDeviceMesh #9

merged 33 commits into from
Sep 28, 2024

Conversation

Jackmin801
Copy link
Member

@Jackmin801 Jackmin801 commented Sep 25, 2024 •
edited
Loading

  • Patch into simulate-multi-node.sh
  • Fix sync

@Jackmin801 Jackmin801 force-pushed the feat-edm branch 3 times, most recently from b09225b to d7532fa Compare September 26, 2024 00:30
Jackmin801
Jackmin801 commented Sep 27, 2024
View reviewed changes
src/zeroband/diloco.py Outdated Show resolved Hide resolved
Jackmin801
Jackmin801 commented Sep 27, 2024
View reviewed changes
src/zeroband/diloco.py Outdated Show resolved Hide resolved
Jackmin801
Jackmin801 commented Sep 27, 2024
View reviewed changes
src/zeroband/diloco.py Outdated Show resolved Hide resolved
Jackmin801
Jackmin801 commented Sep 27, 2024
View reviewed changes
src/zeroband/testing.py Outdated Show resolved Hide resolved
Jackmin801
Jackmin801 commented Sep 27, 2024
View reviewed changes
src/zeroband/train.py Outdated Show resolved Hide resolved
Jackmin801
Jackmin801 commented Sep 27, 2024
View reviewed changes
src/zeroband/train.py Outdated Show resolved Hide resolved
Jackmin801
Jackmin801 commented Sep 27, 2024
View reviewed changes
src/zeroband/comms.py Outdated Show resolved Hide resolved
@Jackmin801 Jackmin801 marked this pull request as ready for review September 27, 2024 05:24
samsja
samsja reviewed Sep 27, 2024
View reviewed changes
Copy link
Collaborator

@samsja samsja left a comment

Choose a reason for hiding this comment

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

I added some comments, looks good, looking forward to try !!

It could be worth to add a small tests, either torchrun one or distributed one.

Would need to test gloo over vpn asap as well

src/zeroband/comms.py Outdated Show resolved Hide resolved
src/zeroband/comms.py Outdated Show resolved Hide resolved
src/zeroband/comms.py Outdated Show resolved Hide resolved
src/zeroband/comms.py Outdated Show resolved Hide resolved
src/zeroband/comms.py Outdated Show resolved Hide resolved
src/zeroband/comms.py Outdated Show resolved Hide resolved
src/zeroband/train.py Outdated Show resolved Hide resolved
src/zeroband/train.py Outdated Show resolved Hide resolved
scripts/simulate_multi_node_diloco.sh Outdated Show resolved Hide resolved
src/zeroband/testing.py Outdated Show resolved Hide resolved
samsja
samsja reviewed Sep 27, 2024
View reviewed changes
Copy link
Collaborator

@samsja samsja left a comment

Choose a reason for hiding this comment

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

you did not install pre-commit hook haha 😿

samsja
samsja reviewed Sep 27, 2024
View reviewed changes
Copy link
Collaborator

@samsja samsja left a comment

Choose a reason for hiding this comment

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

would be nice if you add your github comments as code comments

samsja
samsja reviewed Sep 28, 2024
View reviewed changes
src/zeroband/data.py Outdated Show resolved Hide resolved
samsja
samsja reviewed Sep 28, 2024
View reviewed changes
src/zeroband/data.py Outdated Show resolved Hide resolved
samsja
samsja reviewed Sep 28, 2024
View reviewed changes
src/zeroband/diloco.py Outdated Show resolved Hide resolved
samsja
samsja reviewed Sep 28, 2024
View reviewed changes
src/zeroband/diloco.py Outdated Show resolved Hide resolved
samsja
samsja reviewed Sep 28, 2024
View reviewed changes
src/zeroband/diloco.py Outdated Show resolved Hide resolved
samsja
samsja reviewed Sep 28, 2024
View reviewed changes
src/zeroband/diloco.py Outdated Show resolved Hide resolved
samsja
samsja reviewed Sep 28, 2024
View reviewed changes
Copy link
Collaborator

@samsja samsja left a comment

Choose a reason for hiding this comment

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

I got a bit confused with the FSDP.summon_full but I think that I got it now with the shared memory

samsja
samsja approved these changes Sep 28, 2024
View reviewed changes
Copy link
Collaborator

@samsja samsja left a comment

Choose a reason for hiding this comment

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

LFGTM !

@samsja samsja merged commit 879828a into main Sep 28, 2024
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@samsja samsja samsja approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@Jackmin801 @samsja