Credit-0: terminal reward across exact policy trajectories #8

Merged
perishadmin merged 1 commit from land/emlearn-credit-zero into main 2026-08-03 18:53:31 +00:00
Contributor

Adds a sixty-parameter exact EML trajectory policy, bounded reward-weighted batch credit from one whole-system return, nine-seed certification, frozen uniform ablation, CLI evidence, and explicit factorization limits.

Adds a sixty-parameter exact EML trajectory policy, bounded reward-weighted batch credit from one whole-system return, nine-seed certification, frozen uniform ablation, CLI evidence, and explicit factorization limits.
Credit-0: terminal reward across exact policy trajectories
All checks were successful
guard / guard (pull_request) Successful in 4m4s
guard / guard (push) Successful in 3m51s
6a2655b69d
Adds a sixty-parameter exact EML trajectory policy, bounded reward-weighted batch credit from one whole-system return, nine-seed certification, frozen uniform ablation, CLI evidence, and explicit factorization limits.

Land-Source: emlearn-credit-zero@b70412ea248f61ec55e5c64523ea401e2527f14f
Sign in to join this conversation.
No reviewers
No labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
PerishLab/emlearn!8
No description provided.