Skip to content

Revert hyperparam defaults and eval cadence #4485

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
Merged
relh merged 39 commits into from
Dec 23, 2025
Merged

Revert hyperparam defaults and eval cadence #4485

relh merged 39 commits into from
Dec 23, 2025

Conversation

@relh
Copy link
Contributor

@relh relh commented Dec 22, 2025 •
edited
Loading

Summary

  • revert PPO/trainer defaults and ViT arch defaults
  • keep BPTT/bs overrides and restore total_timesteps=10B
  • set eval cadence to every 150 epochs in cogs_v_clips and machina_1

Testing

  • not run (not requested)

This was referenced Dec 22, 2025
Merged
Merged
@relh Graphite App
Copy link
Contributor Author

relh commented Dec 22, 2025 •
edited
Loading

This stack of pull requests is managed by Graphite. Learn more about stacking.

@relh relh mentioned this pull request Dec 22, 2025
Merged
@relh relh changed the title reset default hyperparms Revert hyperparam defaults and eval cadence Dec 22, 2025
@datadog-official
Copy link

datadog-official bot commented Dec 22, 2025 •
edited
Loading

✅ Tests

🎉 All green!

❄️ No new flaky tests detected
🧪 All tests passed

This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 51b7342 | Docs | Was this helpful? Give us feedback!

@relh relh marked this pull request as ready for review December 22, 2025 16:13
@relh relh force-pushed the rb-stack-4-hyperparam-revert branch from ff1dabc to d240579 Compare December 22, 2025 16:32
@relh relh force-pushed the rb-stack-3-teacher-behavior branch from 00a9d5d to b6e43f1 Compare December 22, 2025 16:32
@relh relh force-pushed the rb-stack-4-hyperparam-revert branch from d240579 to 7e3d145 Compare December 22, 2025 16:41
@relh relh force-pushed the rb-stack-3-teacher-behavior branch from b6e43f1 to 74192f7 Compare December 22, 2025 16:41
@relh relh assigned subho406 and unassigned relh Dec 22, 2025
@relh relh force-pushed the rb-stack-4-hyperparam-revert branch from 7e3d145 to 6503092 Compare December 22, 2025 16:48
@relh relh force-pushed the rb-stack-3-teacher-behavior branch from 74192f7 to 6bd8264 Compare December 22, 2025 16:48
@relh relh force-pushed the rb-stack-3-teacher-behavior branch from 1e1f2c6 to e664ca5 Compare December 22, 2025 19:54
@relh relh force-pushed the rb-stack-4-hyperparam-revert branch from 2efe6db to 028cc3a Compare December 22, 2025 19:54
@relh relh force-pushed the rb-stack-4-hyperparam-revert branch from 028cc3a to d4b372d Compare December 22, 2025 20:02
Base automatically changed from rb-stack-3-teacher-behavior to main December 22, 2025 20:18
@relh relh added this pull request to the merge queue Dec 23, 2025
Merged via the queue into main with commit dfa545e Dec 23, 2025
18 checks passed
@relh relh deleted the rb-stack-4-hyperparam-revert branch December 23, 2025 17:34
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@subho406 subho406 subho406 approved these changes

Assignees

@subho406 subho406

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@relh @subho406 @gustofied