Skip to content

feat(checkpoint): union sharded generation cuts - #3756

Draft
macandro96 wants to merge 1 commit into
amahishi/gym-checkpoint-scale-optimizationsfrom
amahishi/gym-sharded-prefix-recovery
Draft

macandro96 wants to merge 1 commit into
amahishi/gym-checkpoint-scale-optimizationsfrom
amahishi/gym-sharded-prefix-recovery

Conversation

@macandro96

Copy link
Copy Markdown
Contributor

Summary

  • export digest-bound generation-cut index fragments from each policy-model proxy
  • accept the union of peer cut indexes in the authoritative model-ledger commit
  • validate that sharded prefix checkpoints cover every prepared proxy

Stack

Validation

  • git diff --check passed
  • the two-shard prefix-recovery functional test remains to run with the RL counterpart

@copy-pr-bot

copy-pr-bot Bot commented Sep 29, 2026

Copy link
Copy Markdown

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

Signed-off-by: Anish Mahishi <amahishi@nvidia.com>
@macandro96
macandro96 force-pushed the amahishi/gym-checkpoint-scale-optimizations branch from e95d57f to 7645842 Compare September 29, 2026 04:40
@macandro96
macandro96 force-pushed the amahishi/gym-sharded-prefix-recovery branch from 92a5b2c to 0c87a59 Compare September 29, 2026 04:40

This branch has not been deployed

No deployments
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.

1 participant