AMR-SD adds a reflection bottleneck to compress diagnostic signals into self-generated hints and uses asymmetric Causal Information Gain to create sparse token-level advantage signals, outperforming baselines and preventing late-stage collapse in RLVR.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
AMR-SD: Asymmetric Meta-Reflective Self-Distillation for Token-Level Credit Assignment
AMR-SD adds a reflection bottleneck to compress diagnostic signals into self-generated hints and uses asymmetric Causal Information Gain to create sparse token-level advantage signals, outperforming baselines and preventing late-stage collapse in RLVR.