Large reasoning models have recently demonstrated emergent “aha moments” that enable reflective reasoning. However, a non-trivial portion of reflective steps are “low-utility,” where reasoning errors are not corrected, or even correct reasoning trajectories are shifted toward incorrect ones. To address low-utility reflective continuation, this paper proposes RUAR, a utility-guided advantage rescaling method for reflective reasoning that estimates step-level utility from changes in correct-answer probability, without human process labels or a separately learned process reward model. RUAR augments outcome-supervised RL by rescaling the token-level advantages within each reflective reasoning step according to the estimated utility of the reflective continuation, rather than applying a uniform trajectory-level advantage across the reasoning trace. The utility is estimated through answer-forcing utility estimation, which measures how the model’s correct-answer probability changes before and after the reflective step by forcing answer generation from both reasoning states using answer-forcing prompts. Furthermore, RUAR introduces an overthinking-mitigating rescaling mechanism, where utility-guided rescaling is no longer applied from the earliest answer-ready reflection step onward. Experiments demonstrate that RUAR improves the accuracy–efficiency trade-off, substantially reducing generated token usage while often preserving or improving final-answer accuracy.

Existing in-context editing (ICE) methods for multi-hop knowledge editing commonly suffer from paraphrase sensitivity, which refers to the phenomenon where these methods are not sufficiently robust to paraphrased multi-hop questions. To improve retrieval accuracy and knowledge routing to address paraphrase sensitivity, this paper proposes a novel entity-aware inference-time knowledge routing method, referred to as EAIR, which consists of four major steps: 1) Entity-referential query decomposition, which decomposes the original question into multiple entity-referential sub-question instructions; 2) Entity-aware retrieval, which leverages the previously reference-resolved topic entity in the retrieval step; 3) Evidence-conditioned contrastive decoding, which discourages the model from relying on its parametric knowledge and steers the model toward following retrieved edits; 4) Reflection-based knowledge routing, which additionally filters decoding results using refusal-style reflection to mitigate the risk introduced by contrastive decoding. Experimental results across the MQuAKE benchmark family and model scales show that EAIR achieves the highest strict case accuracy in 11 of 12 settings, substantially reducing paraphrase sensitivity.

Masked diffusion language models (MDLMs) enable efficient parallel decoding but are limited by a monotonic unmasking policy, where committed tokens cannot be revised. While remasking-based methods mitigate early errors, they mainly intervene during generation. In this work, we study post-hoc refinement of a completed draft and find that naive correction often fails because of contextual lock-in, a phenomenon in which local error patterns become self-reinforcing. To address this, we propose PURE (Post-hoc Unlocking and REfinement), a training-free inference algorithm for two-phase decoding. PURE profiles confidence dynamics during drafting to identify unstable regions via an instability score (𝛥_i), then unlocks them through deterministic window masking and stochastic leftward relaxation. On reasoning benchmarks, PURE substantially improves accuracy when applied to LLaDA-8B-Instruct, including a gain of +12.9 points over the baseline on GSM8K. These gains require only a small refinement budget, yielding a favorable compute-quality trade-off for discrete diffusion decoding.

Scope ambiguity arises when a single surface string admits multiple logical interpretations depending on how operators such as negation and indefinites interact. Prior work (Kamath et al. 2024) primarily evaluates scope sensitivity through output-level behavior, leaving open whether alternative interpretations are internally represented or merely reflected in shifted token probabilities.

We introduce **an evidence-first, ambiguity-following diagnostic** that asks a more specific question: when the surface string is held constant and only contextual evidence biases interpretation, do transformer language models form distinct internal representations corresponding to alternative logical readings?

To investigate this, we present **SCOPEx**, extending the dataset of Kamath et al. (2024) by systematically manipulating preceding context while keeping the ambiguous sentence identical. We further introduce passivized variants as syntactically less ambiguous controls. Using SCOPEx, we show that transformer LMs:

(i) exhibit **systematic scope-sensitive behavior conditioned on context**, and

(ii) form **separable hidden-state representations for alternative scope readings of the same surface string**.

This divergence emerges in later layers and causally influences downstream predictions.

Standard semi-autoregressive decoding for full-attention masked diffusion language models (MDLMs) computes predictions for future blocks but discards them until those blocks become active. When introduced as soft-future context, these predictions reveal a stability contrast: hypotheses later revised during decoding exhibit larger distributional shifts than those eventually retained. We introduce FUSA (Future-aware Semi-Autoregressive Decoding), a training-free decoder that exploits this contrast by alternating the next block between masked and soft representations and committing active-block tokens that remain stable across the two views, without additional model forward calls. On LLaDA-8B-Instruct, FUSA improves scores across reasoning and code-generation benchmarks while delivering 2.10– 3.94× wall-clock speedups over Top-1 decoding, with gains of up to 5.1 points. Additional results show that FUSA transfers to other MDLMs and remains effective under KV-cached inference and in long-form instruction following. Our code is available at https://github.com/yangryeol/FUSA.