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.

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

Configs for LUMI ablations #324

Draft
wants to merge 8 commits into
base: main
Choose a base branch
from
Draft

Configs for LUMI ablations #324

wants to merge 8 commits into from

Conversation

epwalsh
Copy link
Member

@epwalsh epwalsh commented Oct 10, 2023

We want to run the following models on LUMI with the v1.5 mix:

  • mitch-ish
    ↪ submit with: sbatch scripts/sbatch-128.sh configs/v1_5-mix-medium-mitch-ish.yaml
  • @dirkgr's run without weight tying - not sure where this config is
  • Llama - waiting on Llama Config #317

#SBATCH --job-name=olmo-medium
#SBATCH --account=project_462000229
#SBATCH --output=/pfs/lustref1/flash/project_462000229/logs/%j.log
#SBATCH --nodes=128 # Total number of nodes
Copy link
Contributor

Choose a reason for hiding this comment

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

doubling this?

Copy link
Member Author

Choose a reason for hiding this comment

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

I'm going to test that out first. If it works I'll update.

Copy link
Member

Choose a reason for hiding this comment

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

You can't double that unless you set a micro batch of 1.

@ibeltagy
Copy link
Contributor

Dirk's config is this branch https://github.com/allenai/LLM/tree/DirksRun2

@epwalsh
Copy link
Member Author

epwalsh commented Oct 10, 2023

@dirkgr can you point me to the exact config you were running from this branch? https://github.com/allenai/LLM/tree/DirksRun2

@dirkgr
Copy link
Member

dirkgr commented Oct 10, 2023

I used this script: https://github.com/allenai/LLM/blob/2d4b62d6978e869f700a15268fa6302d78718a06/scripts/v1-mix-medium-on-lumi.sh

The command line was just sbatch scripts/v1-mix-medium-on-lumi.sh --load_path=<whatever_the_latest_run_without_spike_was>.

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.

None yet

3 participants