-
Notifications
You must be signed in to change notification settings - Fork 1
Scaled evals #159
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
base: main
Are you sure you want to change the base?
Scaled evals #159
Conversation
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.
Pull Request Overview
This PR scales evaluations by updating dependencies and modifying the deepspeed accelerate configuration for vllm-based eval scripts.
- Added additional dependencies (setuptools, ipykernel, matplotlib) in pyproject.toml.
- Increased num_processes and added main_process_port in the deepspeed configuration file.
Reviewed Changes
Copilot reviewed 3 out of 5 changed files in this pull request and generated no comments.
| File | Description |
|---|---|
| pyproject.toml | Added dependencies to support additional functionality. |
| benchmarks/ppo/accelerate_configs/deepspeed_zero2.yaml | Updated parallel process count and added a process port setting. |
Files not reviewed (2)
- jobs/validate_all_static.sh: Language not supported
- jobs/validate_all_static_diversity.sh: Language not supported
jobs/validate_all_static.sh
Outdated
|
|
||
| # list all sub‐tasks | ||
| tasks=( | ||
| ultra-hh-sampled |
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.
Why don't we no longer need the full lest of datasets as it was before? The loop is much better, just trying to understand why the change
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.
Should these changes be propogated to the other benchmark deepspeed configs?
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.
Are the dependency changes temporary?
Only a few new scripts to run eval with vllm.