← back to paper

Review history

arxiv: 2608.00536 · 2 revisions

DocPO: Advancing Document Policy Optimization via Tailored Step-Aware Rewards

1. 2026-08-07 CONDITIONAL MODERATE v1.4.0-alltime-deepseek-medium novelty 6.0
   140762 ms 18247 in 12337 out 2026-08-07T00:54:08.479614+00:00
2. 2026-08-05 CONDITIONAL HIGH v1.3.0-alltime-deepseek novelty 6.0
   171577 ms 20651 in 17682 out 2026-08-05T00:41:37.718256+00:00