CORTEXA
← Browse
arxivcs.LGcs.CL2026-06-28

Do Models Read What They Write? Causal Registers in Scratchpad Reasoning

Benjamin Shih, John Winnicki, Eric Darve

A central hope behind process supervision is that models can expose intermediate variables that matter for their later behavior. For this to help with alignment, a scratchpad must be tied to the computation: when the model writes a state, later steps should compute from that state. To test this requirement, we use a controlled state-tracking task with a known update rule, comparing models trained to report only the final state with models trained to write intermediate states before giving the final answer. At evaluation, we edit the internal representation of one written state while leaving the visible scratchpad text fixed. Because the transition rule is known, the edit has a single correct downstream consequence. In Qwen2.5-Coder-7B, the state-writing model predicts the next phase bit implied by the edited state on 80% and 91% of held-out examples across the two task variants, while pretrained and final-answer-only controls remain near baseline. Additional controls rule out generic next-token steering and copying another continuation: the prediction depends on both the edited state and the current move. The same causal-use pattern replicates across model families. Together, these results suggest a sharper goal for scratchpad oversight: not just to make intermediate reasoning legible, but to train written states that the model uses as part of its computation.

View free PDFSource page

Related papers

arxivcs.LGcs.CL2026-07-16

Leveraging Instruction Tuning and Merging for Reasoning Model Adaptation

Yu-Du Feng, Niels Mündler-Sasahara, Mark Vero, Martin Vechev

Reasoning language models (RLMs) have demonstrated impressive performance in domains such as mathematics and coding. These domains permit reliable verification of model outputs, which is important for enabling the reinforcement learning that drives RLM performance gains. However,…

View free PDFSource page
arxivcs.AIcs.CLcs.LG2026-06-29

When Does Learning to Stop Help? A Cost-Aware Study of Early Exits in Reasoning Models

Zhe Dong, Fang Qin, Manish Shah

Reasoning models spend test-time compute unevenly across instances, and a growing family of early-exit rules -- confidence thresholds, entropy monitors, answer-stability checks, and learned stoppers -- promises to reclaim the waste. These rules, however, are evaluated under heter…

View free PDFSource page
arxivcs.CLcs.LG2026-07-07

Mitigating Factual Hallucination in Large Reasoning Models via Mixed-Mode Advantage Regularization

Kaishen Wang, Tong Zheng, Xuehao Cui, Ruibo Chen, Tianyi Xiong, Heng Huang

Large reasoning models (LRMs) improve language model capabilities by generating explicit thinking traces before final answers. In factuality-oriented question answering (QA), such thinking often improves overall performance by helping the model recover relevant knowledge and refi…

View free PDFSource page
arxivcs.LGcs.AIcs.CLcs.MA2026-07-20

MADA-RL: Multi-Agent Debate-Aware Reinforcement Learning for Parameter-Efficient Reasoning in Compact Models

Martino M. L. Pulici, Cuong Xuan Chu, Evgeny Kharlamov, Zifeng Ding, Volker Tresp, Yunpu Ma

Large language models achieve strong reasoning performance, but often at prohibitive training cost - a challenge that is especially acute for compact models ($\leq 4 \, \mathrm{B}$ parameters) trained under limited budgets. We introduce MADA-RL, a post-training framework that spe…

View free PDFSource page