# Study finds most RLVR out-of-domain gains come from answer formatting

> When re‑scored with a neutral extractor, 82% of the reported boost disappears, showing it was a formatting effect.

Oossa · 2026-10-16 · https://oossa.com/en/study-finds-most-rlvr-out-of-domain-gains-come-from-answer-formatting

Researchers Zongye Lyu and colleagues examined reinforcement learning with verifiable rewards (RLVR). They compared a base 7‑billion‑parameter model with its RL‑trained version on four released pairs. Using their own extractor that looks for answers inside a \boxed{} tag, they saw 82% of the out‑of‑domain multiple‑choice improvement reported by public extractors vanish. The gain was mostly the model moving its answer into the box, not answering better.

## The facts

- 82% of the reported gain disappears when using the new extractor
- Study released on 2026-10-16

## Why it matters

It shows that claimed performance boosts from RLVR may be artifacts of how answers are parsed, not real improvements in reasoning.

## Sources & references

1. [Same Answer, New Box: How Much of RLVR's Out-of-Domain Gain Is Answer Format?](https://arxiv.org/abs/2610.20639) – arXiv cs.LG, 2026-10-16

Last updated: 2026-10-16