Correcting What You Cannot See: Credit Assignment for Perception Distillation in Multimodal Reasoners
- Published
- Source
- arXiv
- Paper number
- 781
- Field
- AI / General
- arXiv ID
- 2607.28336
Key points
- It formalizes the credit-assignment problem in multimodal distillation by separating perception failures from reasoning failures.
- A soft AND gate, formed by multiplying the sub-failure rate, or 1-PSR, and the teacher-student mismatch, or KL, increases correction strength only when both signals are present.
- It separates perception and reasoning rollouts, distilling only the perception segment while leaving the reasoning segment to RL.
- When moving from 32B to 8B, the average score across 8 benchmarks rises from 56.94 to 61.22, and when moving from 8B to 2B it rises from 44.50 to 47.28.
- Removing the PCD weight lowers performance by 2.22 points, and removing the separated rollout lowers it by 0.88 points, confirming the ablation.
Paper links
External research summaries. These are not HDATF publications or measured product results.