Privileged-context on-policy self-distillation degrades thinking models' long-budget accuracy by suppressing forking and self-correction behaviors, while helping instruction-tuned models.
Reuse your flops: Scaling rl on hard problems by conditioning on very off-policy prefixes
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
years
2026 6roles
background 1polarities
background 1representative citing papers
VPD frames language feedback learning as variational EM so the teacher policy refines itself via trust-region updates on outcomes while the student learns dense token distributions on its own rollouts, outperforming fixed-teacher baselines on reasoning and code tasks.
AdaPrefix-GRPO treats solution-prefix length as a feedback controller targeting 50% rollout success rate during GRPO training, then anneals to zero prefix, yielding 1.6–2.1× accuracy gains over vanilla GRPO at matched compute on hard math.
DenoiseRL injects wrong reasoning prefixes from weak models into RL rollouts and trains the policy to recover, improving GRPO/DAPO mathematical reasoning on named benchmarks.
REFT improves Pass@1/8/64 in RLVR by uniform first-token sampling from top-N candidates across 0.5B-7B models and multiple difficulty levels.
Prefix Sampling replays self-generated trajectory prefixes to control rollout pass rates near 50% in binary-reward RL, delivering wall-clock speedups and modest performance gains on SWE-bench Verified and AIME tasks.
citing papers explorer
-
Rethinking On-Policy Self-Distillation for Thinking Models
Privileged-context on-policy self-distillation degrades thinking models' long-budget accuracy by suppressing forking and self-correction behaviors, while helping instruction-tuned models.
-
Learning from Language Feedback via Variational Policy Distillation
VPD frames language feedback learning as variational EM so the teacher policy refines itself via trust-region updates on outcomes while the student learns dense token distributions on its own rollouts, outperforming fixed-teacher baselines on reasoning and code tasks.
-
Max Out GRPO Signal: Adaptive Trace Prefix Control for Hard Reasoning Problems
AdaPrefix-GRPO treats solution-prefix length as a feedback controller targeting 50% rollout success rate during GRPO training, then anneals to zero prefix, yielding 1.6–2.1× accuracy gains over vanilla GRPO at matched compute on hard math.
-
DenoiseRL: Bootstrapping Reasoning Models to Recover from Noisy Prefixes
DenoiseRL injects wrong reasoning prefixes from weak models into RL rollouts and trains the policy to recover, improving GRPO/DAPO mathematical reasoning on named benchmarks.
-
Where Rollouts Begin: Low-Load, High-Leverage First-Token Diversification for RLVR
REFT improves Pass@1/8/64 in RLVR by uniform first-token sampling from top-N candidates across 0.5B-7B models and multiple difficulty levels.
-
Rollout Pass-Rate Control: Steering Binary-Reward RL Toward Its Most Informative Regime
Prefix Sampling replays self-generated trajectory prefixes to control rollout pass rates near 50% in binary-reward RL, delivering wall-clock speedups and modest performance gains on SWE-bench Verified and AIME tasks.