PACED applies student pass-rate weighting w(p)=p(1-p) to distillation, concentrating on the zone of proximal development and delivering up to +8.2 gains on AIME tasks with reduced forgetting.
Overconfident errors need stronger correction: Asymmetric confidence penalties for reinforcement learning.CoRR, abs/2602.21420
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
citation-role summary
other 1
citation-polarity summary
years
2026 5roles
other 1polarities
unclear 1representative citing papers
Experiments on coding and deterministic tasks demonstrate that data gating is sufficient for self-play stability while reward variants are not, revealing the Grounded Proposer Paradox and a two-stage phase transition under continuous gate strictness.
A two-axis taxonomy of student entropy and teacher-student divergence identifies informative tokens in on-policy distillation, allowing near-full performance with 10-50% of tokens.
CRISP achieves 57-59% token reduction on MATH-500 with 9-16 point accuracy gains on Qwen3 models via iterative self-distillation of concise reasoning behavior.
Mixed-Policy Distillation transfers concise reasoning behavior from larger to smaller LLMs by having the teacher compress student-generated trajectories, cutting token usage up to 27% while raising benchmark scores.
citing papers explorer
-
PACED: Distillation and On-Policy Self-Distillation at the Frontier of Student Competence
PACED applies student pass-rate weighting w(p)=p(1-p) to distillation, concentrating on the zone of proximal development and delivering up to +8.2 gains on AIME tasks with reduced forgetting.
-
Survive or Collapse: The Asymmetric Roles of Data Gating and Reward Grounding in Self-Play RL
Experiments on coding and deterministic tasks demonstrate that data gating is sufficient for self-play stability while reward variants are not, revealing the Grounded Proposer Paradox and a two-stage phase transition under continuous gate strictness.
-
TIP: Token Importance in On-Policy Distillation
A two-axis taxonomy of student entropy and teacher-student divergence identifies informative tokens in on-policy distillation, allowing near-full performance with 10-50% of tokens.
-
CRISP: Compressed Reasoning via Iterative Self-Policy Distillation
CRISP achieves 57-59% token reduction on MATH-500 with 9-16 point accuracy gains on Qwen3 models via iterative self-distillation of concise reasoning behavior.
-
Reasoning Compression with Mixed-Policy Distillation
Mixed-Policy Distillation transfers concise reasoning behavior from larger to smaller LLMs by having the teacher compress student-generated trajectories, cutting token usage up to 27% while raising benchmark scores.