PDCR improves vision-language reasoning by computing separate normalized confidence advantages for perception steps and reasoning steps after unsupervised decomposition.
Beyond the first error: Process reward models for reflective mathematical reasoning
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
years
2026 4representative citing papers
A trajectory-aware process reward using DTW on sentence embeddings, combined with exact-match in GRPO after SFT, raises mean medical VQA accuracy from 0.598 to 0.689 across six benchmarks.
PROGRS uses outcome-conditioned centering on PRM scores to safely integrate process rewards into GRPO for improved Pass@1 on math benchmarks.
citing papers explorer
-
PDCR: Perception-Decomposed Confidence Reward for Vision-Language Reasoning
PDCR improves vision-language reasoning by computing separate normalized confidence advantages for perception steps and reasoning steps after unsupervised decomposition.
-
Improving Medical VQA through Trajectory-Aware Process Supervision
A trajectory-aware process reward using DTW on sentence embeddings, combined with exact-match in GRPO after SFT, raises mean medical VQA accuracy from 0.598 to 0.689 across six benchmarks.
-
LLM Reasoning with Process Rewards for Outcome-Guided Steps
PROGRS uses outcome-conditioned centering on PRM scores to safely integrate process rewards into GRPO for improved Pass@1 on math benchmarks.
- Internalizing Outcome Supervision into Process Supervision: A New Paradigm for Reinforcement Learning for Reasoning