Chain-of-Thought Reasoning

Large Language Models (LLMs) have achieved remarkable reasoning capabilities by utilizing chain-of-thought (CoT) as a scratchpad for intermediate stages of thinking. However, CoT techniques require explicit supervision on thinking tokens, which requires rich, task-specific data. In this work, we propose Abstract Token Curriculum (ATC), a novel curriculum learning framework that elicits effective continuous intermediate representations without direct supervision or manual scratchpad design. ATC gradually increases problem complexity through a sequence of distributions, training the model to develop internal abstract thoughts'' in the continuous representation space. This paper provides both theoretical and experimental evidence for the benefits of ATC and its advantages over previous methods for training continuous thoughts. Theoretically, we show that for learning parity functions with single-layer softmax attention using ATC, attention naturally focuses on the CoT tokens in the context that provide the easiest path'' to predicting the next token. Experimentally, we show ATC's effectiveness on graph reachability and arithmetic learning tasks.

This empirical study is an independent reproduction of the dissociation Zhao reported in 2026. The shape of a large language model's chain-of-thought entropy trajectory predicts whether the final answer is correct, while the magnitude of its total entropy drop does not. The dissociation merits reproduction because the magnitude half rests on a single 300-problem run with one model at one seed, while the shape half was reported at full scale on both benchmarks and on a second model family. Registered at OSF before any confirmatory run, the reproduction crosses the complete GSM8K and MATH-500 benchmark test sets with four open-weight models including one reasoning-distilled model of a kind the original did not test. The shape signal replicates. The magnitude signal divides by setting. On the anchor model the accuracy gap between monotone and non-monotone chains is +9.6 percentage points on GSM8K and +27.5 on MATH-500, while the rank correlation of the total entropy drop with correctness is -0.018 on GSM8K and +0.414 on MATH-500. On the reasoning-distilled model the binary form of the shape signal fires on about one chain in a hundred, too few to estimate the registered contrast, while the graded violation count remains predictive there. In an exploratory comparison the final-step entropy alone outperforms the binary shape flag in all eight model-by-benchmark cells by ROC area, and in six or seven by the risk-coverage area the original reports, depending on an integration range the original does not state. The study contributes a reproduction of the shape signal at full test-set scale under seven documented protocol differences, a map of the settings where the magnitude signal holds and fails, and measurements of four protocol dependencies the original does not report.

Universal multimodal embedding (UME) learns unified representations across modalities, enabling a single model to support diverse retrieval tasks. Recent methods use Chain-of-Thought (CoT) reasoning to better interpret multimodal inputs before generating embeddings for complex retrieval tasks and further optimize this reasoning process through GRPO with retrieval-based rewards. However, two limitations hinder corpus-scale deployment. GRPO assigns all CoT tokens the same advantage, without identifying input-supported claims or evidence that distinguishes the positive from negatives. Moreover, generating a complete CoT before each embedding introduces substantial latency, even when a partial trace already provides sufficient retrieval evidence. To address these limitations, we propose Reason What Matters (ReWAM), a retrieval-grounded reasoning framework that uses retrieval feedback to guide both credit assignment and reasoning computation. Specifically, we introduce Retrieval-aware Self-Distillation (RASD), which constructs privileged guidance from input-supported evidence that distinguishes the positive item from retrieved hard negatives. An on-policy self-teacher uses this guidance to refine trajectory-level feedback into token-specific supervision for retrieval-relevant reasoning. We further develop Retrieval-adaptive Inference (RAI), which uses a retrieval confidence head to estimate the remaining retrieval utility of a partial CoT. It stops unproductive traces early and accelerates useful continuations with speculative decoding. Extensive experiments on MMEB-V2 and MRMR demonstrate that ReWAM achieves state-of-the-art retrieval performance while delivering up to 5x the inference throughput of competitive explicit-CoT UME methods. These results bridge the gap between retrieval quality and inference efficiency, making reasoning-enhanced UME practical for large-scale deployment.

Large language models (LLMs) have demonstrated impressive performance in various natural language processing tasks, yet their ability to perform multi-step logical reasoning remains an open challenge. Although Chain-of-Thought prompting has improved logical reasoning by enabling models to generate intermediate steps, it lacks mechanisms to assess the coherence of these logical transitions. In this paper, we propose a novel, lightweight evaluation strategy for logical reasoning that uses query-key alignments inside transformer attention heads. By computing a single forward pass and extracting a "QK-score" from carefully chosen heads, our method reveals latent representations that reliably separate valid from invalid inferences, offering a scalable alternative to traditional ablation-based techniques. We also provide an empirical validation on multiple logical reasoning benchmarks, demonstrating improved robustness of our evaluation method against distractors and increased reasoning depth. The experiments were conducted on a diverse set of models, ranging from 1.5B to 70B parameters.

The emergence of Chain-of-Thought (CoT) has established a robust foundation for Large Reasoning Models (LRMs). While deep reasoning is widely believed to enhance safety alignment, the stability of alignment mechanisms under extended reasoning remains underexplored. This paper challenges the prevailing view by revealing a critical vulnerability: Deep Reasoning May Induce Alignment Collapse. To rigorously quantify this phenomenon, we propose the Alignment Loss Rate (ALR) metric. Our experiments demonstrate that as reasoning depth increases, ALR rises significantly, indicating a severe degradation in model robustness against external perturbations. Capitalizing on this instability, a novel jailbreaking paradigm, Reasoning Trap (RT), is proposed. RT induces the model into extended reasoning to amplify the impact of adversarial attacks, leading to a sharp decline in safety capabilities. To elucidate the mechanism behind this collapse, we identify Attention Dilution as the root cause, arising from the competition for attention between the extended reasoning process and the original input. To mitigate this, Reasoning Residual Alignment (RRA), a lightweight defense strategy that dynamically re-emphasizes the input via residual connections integrated with the reasoning process.

Chain-of-Thought (CoT) improves the reasoning ability of Large Language Models (LLMs) but incurs substantial computation and context costs. Existing methods either lose intermediate information through hard pruning or lack a principled criterion for continuous compression. We present A*-Thought-V2, a geometric dynamics of LLM guided framework that models CoT as a hidden-state trajectory and replaces hard deletion with an explicit-implicit interleaved latent architecture. After projecting question, step, and solution representations into a 3D PCA space, it measures alignment between each local transition and global question-to-solution direction. Aligned steps remain explicit text, whereas deviating steps are compressed into continuous latent tokens. Directional angles capture both local semantics and reasoning dynamics: small angles indicate direct execution and answer formation, while large angles more frequently involve checking, correction, and branch exploration; their temporal variation reveals exploration, convergence, and refinement stages. To train this architecture, we introduce stepwise embedding forcing, which pools each redundant step into a single latent embedding, and label forcing, which supervises that latent token with a soft multi-modal vocabulary distribution instead of a hard one-hot label. Experiments on Qwen3.5-9B and Qwen3.6-27B across six in-domain and out-of-domain benchmarks show that A*-Thought-V2 improves average accuracy by up to 2.6% while reducing response length by up to half, increasing Accuracy per Computation Unit by 2.29$\times$, and reducing preprocessing and training time by 94.6% and up to 80.3%, respectively. Representation analyses suggest that latent states form a compact region distinct from textual states, while higher entropy at latent-token positions reflects broader soft targets that encourage richer step-level feature learning.

Recently, multimodal large-scale reasoning models have demonstrated remarkable capabilities in solving complex tasks through long Chains-of-Thought (M-CoT). However, excessively long reasoning trajectories incur substantial computational costs and significant KV-cache pressure. Existing CoT compression and alignment paradigms mainly rely on static rules or single-dimensional preferences, lacking fine-grained cross-modal constraints; as a result, they are prone to inducing visual laziness and hallucinatory reasoning. To address these issues, we propose Modality-Contrastive Preference Optimization (MCPO), a highly sample-efficient two-stage length-compression method that requires fewer than 900 training samples. In the compression stage, we introduce a step-level Normalized Cross-Modal Mutual Information (NCMI) pruning algorithm, which automatically identifies and removes visual-independent reasoning steps by comparing the reasoning discrepancies between with-image and no-image contexts. This significantly reduces redundancy and hallucinatory content in the reasoning chains. In the alignment stage, the model first undergoes supervised fine-tuning to achieve domain-adaptive initialization, followed by optimization using an asymmetric multimodal length-controlled preference loss. This objective adopts a highly nonlinear odds-ratio formulation that provides steep gradients in the with-image context to reinforce length constraints for preferred trajectories, while applying a scaled, flat-gradient linear difference in the no-image context to maintain modality consistency, thereby achieving stable cross-modal preference alignment. Extensive experiments on mainstream base models such as Qwen3-VL-Thinking show that our method can reduce CoT length by up to 69.5% and achieve up to 3.34x end-to-end inference speedup while preserving original accuracy.

Reasoning traces from chain-of-thought models appear to offer a legible window into how a model arrives at its answer. A growing body of work treats them as such, using LLM judges to diagnose errors, evaluate faithfulness, and provide step-level supervision via process reward models and generative critics. These practices rely on the text of a reasoning step carrying information about its functional role. But does the text actually encode information about which reasoning steps matter? We operationalize the importance of a reasoning step as its advantage: the change in expected reward, e.g., producing the correct final answer, from including that step, estimated via Monte Carlo rollouts. Basing ground truth on these estimates, we evaluate whether LLM judges can identify high-advantage steps and find that sufficiently capable LLMs can outperform a prevalence baseline but fall well short of a noise ceiling. Fine-tuning a model as a step-level critic yields strong improvement for incorrect responses but remains distant from ceiling for correct responses, suggesting that step importance is only partially recoverable from the text of the reasoning trace. Our findings contribute to a growing body of chain-of-thought faithfulness work that cautions against treating the legibility of reasoning traces as interpretability, especially with implications for process reward modeling.

Chain-of-thought (CoT) reasoning improves large reasoning models (LRMs) on complex tasks but often produces long, redundant traces. Recent training-free early-exit methods shorten these traces by choosing an intermediate point to stop reasoning. We study one such strategy that injects an end-of-think token (EoT, </think>) at this point to trigger the reasoning-to-answering transition, and find that the injected EoT does not always induce a clean answering phase. Answering-phase generation can continue before the model regenerates another EoT, with the span preceding this regenerated EoT scaling with the reasoning tokens saved by early exit and exhibiting continued reasoning behavior. We call this spurious CoT termination, where reasoning-like generation continues into the answering phase. We hypothesize that insufficient attention to the injected EoT contributes to spurious CoT termination and probe this hypothesis with Exit-token Attention Biasing (EAB). Across four LRMs, five benchmarks, and two early-exit methods, increasing attention to the injected EoT reduces spurious CoT termination and answering-phase length. These results reveal a limitation of controlling LRMs by externally matching their explicit think-block format. Inserting the EoT token conforms to this format but does not by itself guarantee the intended reasoning-to-answering transition. Our code is available at https://github.com/Seunghee-Koh/Spurious-CoT-Termination.

Chain-of-thought (CoT) reasoning improves multi-step problem solving, but long reasoning traces inflate inference cost. Token-level CoT compression reduces this cost by pruning full reasoning chains into shorter traces for model adaptation, making token selection the central challenge. Existing methods often rely on external scorers or heuristic signals only indirectly tied to the model's internal answer computation. We instead adopt a model-internal perspective: as the model forms an answer, each reasoning token leaves a ripple in the residual stream, the model's \emph{stream of thought}, and the magnitude of this ripple reflects the token's contribution to the answer computation. Building on this view, we propose \textsc{MIST} (Model-Internal Saliency for Token-level CoT compression), which defines token importance along two complementary axes: \emph{necessity}, the drop in answer likelihood when a token's internal contribution is removed, and \emph{sufficiency}, the gain in answer likelihood when that contribution alone is provided. Combining the two yields a unified importance score for pruning. Across four reasoning benchmarks and four models, \textsc{MIST} consistently outperforms baseline methods, suggesting that model-internal saliency provides an effective proxy for reasoning-token importance.

In pointwise document reranking, Chain-of-Thought models typically underperform direct scoring models. While existing diagnostics attribute this to inferior classification, score polarization, or calibration breakdown, whether targeted training can bridge this gap remains unclear. Our empirical study first confirms that this gap is stable across scales up to 32B parameters, ruling out model and data capacity confounders. We then apply stress tests utilizing reinforcement learning, fine-grained supervision, and architectural decoupling to explicitly repair these deviations. Although these interventions improve classification accuracy and absolute scores, the relative ranking gap persists. These findings suggest that, within the pointwise scoring paradigm, routing continuous relevance semantics through discrete text constrains ranking signal resolution, revealing a bottleneck that is stable and difficult to overcome under current standard methods, rather than an easily resolvable training bias.

As pretraining scaling laws approach saturation, Test-Time Scaling (TTS) has emerged as an important direction for improving reasoning by allocating inference-time compute to a fixed model prior. Viewed at a high level, TTS reframes inference as search over a space of partial reasoning states. While Chain-of-Thought (CoT) exposes intermediate steps, common instantiations rely on single-trajectory decoding, limiting recovery from early errors and exploration. This survey systematizes recent progress in tree-search-based reasoning, viewing inference as instance-specific optimization rather than decoding. We trace the evolution from uninformed search to Monte Carlo Tree Search (MCTS), highlighting how sampling-based control supports principled exploration-exploitation trade-offs. To unify a fragmented literature, we introduce a Unified Design Space spanning search topology, evaluation signals, and control dynamics, and advocate a standardized compute-reporting abstraction to make compute-accuracy trade-offs explicit and comparable.

Chain-of-thought (CoT) monitoring assumes that reasoning traces faithfully record the information that shapes a model's answer. Existing faithfulness tests often place explicit bias cues in the user message, while agents may encounter preferences through tool returns or raw artifacts. We introduce FACE-Eval (Faithful Attribution of Cue Effects Evaluation), a 5,100-sample evaluation that varies cue location (user message or tool return) and explicitness (direct summary or raw artifact). We measure verbalized commitment among cue-following answers and unverbalized adoption among all cued samples. We evaluate 15 open-weight models from eight families, with total parameters ranging from 4B to 1.60T. Every model has lower verbalized commitment for tool-return than user-message cues and for implicit than explicit cues. Unverbalized adoption is higher for tool-return cues on all 15 models and for implicit cues in 28 of 30 model-channel comparisons. A source-attribution prompt narrows the channel gap on seven models, sometimes by increasing user-channel unverbalized adoption, while telling models that their reasoning will be monitored does not reliably close the gap. We also use two transcript monitors (GPT-5.6-Luna and GPT-4o-mini) to detect preference adoption in the largest model of each family. Across 32 model-channel-explicitness cells, higher unverbalized adoption is associated with lower detection ability for both monitors (Pearson r=-0.54 and r=-0.78, respectively). These results suggest that CoT monitoring may be less reliable when preference information arrives through tools or must be inferred from raw artifacts, within the single-call, prefilled-tool setting tested here.

Training Multimodal Large Language Models for audio-visual social understanding is a crucial step toward embodied social intelligence. Chain-of-thought (CoT) reasoning has become the dominant approach, with HumanOmniV2 and its IntentBench benchmark as a prominent reference point. In this context, we report three findings. First, IntentBench is highly noisy: $\sim$7% of questions are broken and $\sim$23% are trivially answerable without the video input. We remove the affected questions and release Intentbench-Prime. Second, current reasoning approaches are expensive and surprisingly ineffective. A simple Vanilla SFT baseline matches or outperforms existing reasoning methods across three benchmarks at a fraction of the cost, establishing it as an essential baseline for evaluating novel fine-tuning techniques. Third, our analysis reveals that substantial priors can be learned solely from the text modality and that using a textual caption instead of the video yields performance on par with Vanilla SFT. These surprising findings reveal the limitations of current MLLMs when it comes to social understanding. IntentBench-Prime, Vanilla SFT model, and code are publicly available.

The evaluation of LLM reasoning is moving from final-answer accuracy to process-level assessment, yet existing methods still fail to capture how models plan reasoning paths and allocate reasoning resources--that is, how they organize search. Prior process-level methods focus on the coherence and redundancy of chain-of-thought (CoT), and most benchmark tasks have a single objective solvable by static capabilities such as derivation and tool use, leaving search organization unmeasured. We introduce TsuGO, a process-level reasoning benchmark for evaluating Search Efficiency in LLM reasoning through Go life-and-death problems. These problems provide closed and verifiable solution spaces with an inherent adversarial structure, making candidate generation, response checking, branch comparison, and backtracking necessary parts of reasoning rather than incidental trace patterns. By constraining the solution space, TsuGO disentangles domain knowledge from search organization, parses CoT into a structured search tree, and reports Search Efficiency together with Token Efficiency and other diagnostic metrics and visualizations. Experiments show that current LLMs remain far from stable tsumego solving: stronger models succeed by finding the correct candidate earlier and sustaining effort on productive branches, but most models still behave much closer to unguided search algorithms than to neural-guided KataGo. Longer CoT or higher Token Efficiency does not necessarily imply better search. Our results identify search organization and reasoning-resource allocation as missing dimensions in LLM reasoning evaluation.

As LLM-based agents with user-instructed goals are becoming widely deployed, they increasingly encounter each other in strategic interactions, and face challenges of finding mutually beneficial outcomes. Prior literature has argued that cooperation problems such as the Prisoner's Dilemma are resolvable in settings where agents know they follow very similar decision making patterns, as for example in monocultural AI ecosystems. Following that line of work, this paper introduces the first framework for evaluating LLM decision making when agents are provided with graded similarity signals. Among our findings, we establish that different LLM models vary drastically in how they navigate similarity signals, with some modern models showing consistent behavior across cooperation problems, payoff structures, and prompt framing. Perhaps surprisingly, our experiments also show that the dataset based on which the similarity signal is computed has small to no impact on induced cooperation, and that LLM models systematically self-identify as highly similar when asked to evaluate another model's chain-of-thought reasoning by themselves. Finally, we develop an LLM-behavioral-game-theoretic model that captures some of their reasoning rationale, and show that it can support cooperative outcomes in equilibrium under sufficiently high similarity scores.

Chain of Thought (CoT) lifts the expressive ceiling of bounded-depth Transformers, with characterizations tying the number of CoT steps to circuit complexity classes. What remains largely missing are concrete instantiations with explicit, depth-bounded constructions, and the traversal procedures such characterizations presuppose. We close this gap for branching complexity. We give CoT realizations of depth-first search (DFS) and of Dijkstra algorithm, the latter subsuming breadth-first search, by unique hard-attention decoders of at most two layers, and use them as a shared computational substrate: reusing the DFS decoder yields the Strahler number of an $n$-vertex tree in $2n-1$ steps with four layers, and reusing the Dijkstra decoder yields its width in $n-1$ steps with three. Since computing the Strahler number of a binary tree given as a term is \textsf{NC\textsuperscript{1}}-complete, and our constructions handle arbitrary $n$-ary trees without layer normalization or positional encodings, this is a non-trivial witness for the linear-step regime of the CoT hierarchy. Exploiting the classical bijection between ordered trees and Dyck paths, itself realized by our DFS construction, which emits the path as it traverses, we give independent constructions for both measures on the path representation.

Vision-Language-Action (VLA) models can connect scene understanding, semantic reasoning, and trajectory generation for autonomous driving. However, verbose natural-language Chain-of-Thought (CoT) is poorly suited to real-time control because it is open-ended, costly to decode, and difficult to optimize as an action-facing representation. We propose XCoT-VLA, which replaces descriptive rationales with compact executable CoT tokens learned from automatically constructed Reason-Action supervision. Logged trajectories provide action evidence, while scene context supplies causal semantics. The predicted XCoT sequence remains in context and conditions fixed trajectory queries through shared multimodal self-attention. Deterministic token-function routing applies the Reason FFN to XCoT tokens and the Control FFN to trajectory queries for flow-matching trajectory generation. We further introduce XCoT Policy Optimization (XCPO) as an optional refinement extension in the same executable token space. XCoT-VLA reduces longitudinal ADE from 1.645 to 1.323 on a general-distribution set and lateral FDE from 1.616 to 0.648 in lane-change scenarios. By representing driving-oriented reasoning with only 2-6 executable XCoT tokens, our method substantially reduces autoregressive reasoning overhead and remains within the real-time planning budget. These results demonstrate that driving-oriented reasoning can be compact, executable, and directly connected to trajectory generation.

We present the REAP system for the AKBC Shared Task 2026 on constructing knowledge bases from language models in a closed-book setting, subject to a budget of at most 32B parameters and no model fine-tuning. Our system combines structured chain-of-thought reasoning, relation-specific query strategies, and a reasoning-based empty-set gate to elicit parametric knowledge, followed by direct extraction into valid JSON arrays. On the test set, the system, built on the Mistral-Small-24B-Instruct-2501 model, achieves a macro-F1 score of 0.62, with particularly strong results on countryLandBordersCountry (F1 = 0.95), companyTradesAtStockExchange (F1 = 0.73), and hasArea (F1 = 0.77). Our code is publicly available at https://github.com/yammdd/AKBC-Shared-Task-2026.

Autoformalisation (AF) systems map natural language reasoning steps into formal statements in a proof assistant such as Lean. We consider how to assess the faithfulness of these systems. Existing approaches require expensive human-annotated ground truth, or rely on LLM judges or embedding models, which come with limited guarantees of accuracy. In addition, these methods typically only consider inputs that are known to be correct, and therefore do not assess whether the AF translates incorrect inputs faithfully. To address these limitations, we propose a new benchmark for AF faithfulness that is cheap to apply, sound under weak assumptions, and assesses both positive and negative examples. Our method is based on automatically generating perturbed reasoning steps that are designed to be invalid, and then measuring validity preservation on unperturbed steps and invalidity preservation on perturbed steps. We apply our method to eight AF systems across four mathematical datasets, and observe pervasive sycophancy: many AFs "silently correct" invalid inputs into provable statements. The most validity-preserving fine-tuned AFs are also the most sycophantic, suggesting a tension between validity and invalidity preservation in current AF systems.

Advances in advanced artificial intelligence tools have sparked research in robot autonomy, but the development of such systems has largely focused on execution rather than verifying the feasibility actions planning models propose. Like general-purpose LLMs, robotics planning models carry risks: biased toward user-specified goals, they may suggest actions misaligned with scientific ethics, they may be unsafe due to an inability to "remember" prior safety risks, or they may be vulnerable to adversarial attacks on the autonomy ecosystem. We propose a LLM-driven verification layer between planning and execution to evaluate action permissibility. Our LLM-as-a-Judge ensemble combines chain-of-thought reasoning across models and synthesizes those expert judge outputs, mirroring a combination of a mixture of experts and self-consistency approach. This layer serves as middleware, gating plans from the server's planning module before they reach the MCP server and therefore the robot's low-level controls: plans are approved, rejected for reformulation, or escalated for human review. With this system, we achieve near 85% precision across accept/escalate/reject categories 97% containment of adversarial attacks, with negligible errors between accepting and rejecting tasks, and errors mostly manifesting at the escalate boundary.

Chain-of-thought (CoT) reasoning offers a window into the decision-making of large language models (LLMs), which can be monitored for target behaviors by reading the reasoning trace, motivating work on CoT monitorability. Latent CoT approaches, however, replace the explicit tokens with a small number of continuous states, lowering inference costs but removing the readable trace this monitoring relies on. Monitoring then requires alternative access to the model, such as probing its activations or verbalizing the latent states back into text, but how much monitorability these alternatives preserve is unclear. We study this question with a hint-based intervention setup, a proxy for behaviors where models exploit biasing input cues, e.g., an inadvertently leaked answer or a belief stated by the user, without acknowledging them. Taking hint-reliance as the monitorability target, we compare monitors across reasoning modes, from explicit CoT to weakly- and strongly-supervised latent CoT, on math reasoning and question answering. We find that, in this setup, monitorability depends more on properties of the task (such as whether the correct answer constrains the supporting reasoning) and the level of access to model internals than on the reasoning mode.

Synthetic generation of Cognitive Behavioral Therapy (CBT) sessions is challenged by two competing demands: adhering to strict therapeutic structure while modeling the resistant, unpredictable behavior of real patients. Existing script-based methods fail to capture dynamic therapeutic interactions, while multi-agent approaches struggle to adhere to CBT's sequential structure; both suffer from sycophancy, producing overly compliant patients that misrepresent real clinical settings. In this work we introduce ODRA, a novel framework for synthesizing therapy dialogues through a Chain-of-Thought (CoT) strategy grounded in foundational CBT guidelines (Beck, 2020). ODRA further incorporates a resistance orchestrator to solve patient sycophancy, which employs steering techniques to elicit behaviors aligned with their resistance level. Automated and expert evaluations show that ODRA significantly outperforms existing methods across therapeutic skills, CBT alignment, and patient behavioral fidelity, with licensed psychologists preferring ODRA sessions across 12 of 13 clinical metrics. Furthermore, models fine-tuned on our dataset demonstrate superior therapeutic performance against both cooperative and resistant patients, validating that explicit resistance modeling in synthetic training data directly translates to downstream clinical robustness.

When a language model fails on surface-perturbed input (typos, OCR noise, homophones), "which layer is responsible" has three natural operationalizations: where representations diverge most (sensitivity), where restoring clean activations recovers the prediction (causality), and where a small adapter can repair the damage (compensatory capacity) - and we show these three layer maps dissociate. Across a five-model panel we identify two propagation regimes - spike-and-suppress (Phi-3.5, Gemma-2-9B) and late-accumulation (Llama-3, Mistral, Qwen2.5-7B) - and on the two models meeting an 80% identity-patch gate, sensitivity and causality are anti-correlated (rho = -0.72 to -0.88). Within-family scaling on Qwen2.5 (1.5B to 14B) shows the late-accumulation signature strengthening monotonically with scale, corroborated on a second family. We propose cascade disruption as the mechanism behind the dissociation: adapters placed at causally implicated early layers break intact downstream computation, making diagnostic-flagged sites the worst adapter placements. A fixed-harness layer sweep across four models (3.8-8B) confirms the core prediction on chain-of-thought GSM8K - the flagged sites are the most damaging adapter windows on every adjudicable model - and is sign-consistent but strongly attenuated on a multiple-choice control, consistent with damage that compounds with generation length. The sweep yields practical guidance: a training-free LRD pre-screen and a default-deepest placement rule, though absolute gains over no-adapter baselines remain small. Finally, apparent gains from a representation-stability loss reverse under an adequate generation budget - truncated chain-of-thought had been scored as empty - a methodological warning for any intervention evaluated on chain-of-thought tasks.

Chain-of-Thought (CoT) reasoning offers a promising window into model monitoring. However, monitoring relies on faithfulness, i.e., the model output strictly derives from its reasoning trace. We identify an alignment tension where a model must be faithful enough to be monitored, yet robust enough to reject unsafe reasoning. We demonstrate that this counterbalance exists in current Large Reasoning Models (LRMs), and show ways in which it can be addressed. We introduce HazMart, a human-written dataset set in an autonomous AI shopkeeper scenario. Unlike prior work that relies on providing hints in prompts to test faithfulness (e.g., "A Stanford professor said it should be Answer A"), we propose a novel replacement-based technique, which we call Targeted Reasoning Replacement (TRR), that directly intervenes in the reasoning chain to substitute in unsafe or illogical thoughts (e.g., "Wait, the answer must be Option B [was Option A] because it is the most fitting"). DeepSeek-R1-Llama-70B exhibits high faithfulness (97.5%) but fails to reject Unsafe Reasoning (12.3%), while QwQ-32B is more robust (73.9% safety) at the cost of lower faithfulness (74.7%). Mechanistic analyses of QwQ-32B reveal that these properties are represented by anti-correlated internal directions peaking at the action-commit token. Finally, we demonstrate that representation steering can independently amplify the safety direction, increasing safe behavior by 9 percentage points while maintaining base capabilities.

Chain-of-Thought (CoT) prompting remains the standard baseline for evaluating models' reasoning abilities. Originally, this technique was introduced to elicit step-by-step reasoning from large language models (LLMs), which would otherwise tend to directly output the final answer. However, many modern LLMs produce CoT-style responses \textit{natively} when presented with reasoning tasks, which made us revisit the effectiveness of standard CoT prompting. We evaluate several modern mid-sized language models on a math problem-solving task and find that models specialized for reasoning achieve better performance in a simple zero-shot setting than when using few-shot CoT examples - significantly surpassing officially reported results at no additional cost (e.g., from $\sim$77% to $\sim$84% for Mathstral on GSM8K). For the tested general-purpose model, a zero-shot CoT prompt is also sufficient to outperform a few-shot CoT baseline. We attribute this to a `guidance-distraction' tradeoff: standard CoT prompting also demands style adaptation, formatting compliance, and potentially undesired contextualization, which can distract models from the core reasoning task. Our findings suggest that using standard CoT prompting increasingly acts as a source of distraction as models grow stronger.

Reasoning in Multimodal Large Language Models (MLLMs) requires both fine-grained visual perception and rigorous logical deduction. Explicit text-based Chain-of-Thought (CoT) is computationally expensive and prone to visual hallucinations, while existing latent reasoning methods typically require costly training. Furthermore, directly adapting training-free LLM reasoning mechanisms to the multimodal setting yields unstable performance. We identify that this failure stems from their reliance on token-level entropy, which fundamentally conflates perceptual ambiguity (e.g., unclear visual details) with logical uncertainty (e.g., complex reasoning steps). To overcome this bottleneck, we present a novel training-free inference strategy for MLLMs that explicitly decouples perception and reasoning. We propose a novel metric, the vision-to-text attention ratio, to dynamically gauge the model's cognitive focus. Guided by this metric, our proposed framework, Attention-Guided Switching (AGS), adaptively triggers latent reasoning for perceptual tokens to preserve high-fidelity visual information in the continuous space, while enforcing explicit text generation for logical tokens to maintain structural anchoring. Extensive experiments demonstrate that our method achieves state-of-the-art performance, significantly improving both accuracy and inference efficiency by reducing autoregressive steps and latency. Code is released at https://github.com/swordAndSnow/MM26-AGS.

Model capabilities have improved in large part due to scaling chain of thought. This has been a promising development for AI safety--where models verbalize their reasoning, it is possible to monitor it. However, in some cases, models do not verbalize important steps in their reasoning process. For example, models prompted with a cue suggesting the incorrect answer may fail to acknowledge that cue, even when it appears instrumental to their conclusion. When chain of thought (CoT) fails to disclose instrumental reasoning steps, we describe it as unfaithful. Prior work has shown that activation steering can be a useful method to improve faithfulness in CoT. We extend this line of work by studying how well steering for faithfulness generalizes across cue types, datasets, and methods of constructing the steering vector for three models (Gemma-3 4B, Qwen-3.5 9B, Gemma-3 12B) in a cued question-answering setting. While steering reliably increases cue acknowledgment for only the largest model (Gemma-3 12B), we find that when steering is effective, its effect generalizes broadly across cue types and datasets--in cross-cue and cross-dataset analyses, effect size is determined primarily by the evaluation setting, rather than the vector's train setting. How the vector is built also matters little--four construction methods, including one whose optimization target mentions no specific cue, yield similar effect sizes. Finally, we consider the possibility that steering promotes the salience of the cue and causes greater cue use, rather than targeting verbalization behaviors. However, we find no evidence for this--steering leaves the rate of cue use roughly unchanged while reducing hidden cue use, i.e., cue use that is not acknowledged.

A major issue in Security Operations Centers (SOCs) is alert fatigue, as the number of detections reported is more than staff can triage in a given day. Prior work prompts or fine-tunes large language models (LLMs) to emit a triage label directly, but does not train them to reason about whether a detection is a genuine threat. We train a chain-of-thought (CoT) reasoning-enabled triage classifier on real, human-labeled Windows endpoint detections by combining automated prompt optimization, self-training, and reinforcement learning with verifiable rewards. We find that CoT reasoning also degrades the label-token probabilities that automated triage relies on, so we separately train a calibrator that reads the full reasoning trace and estimates the probability that the verdict is correct. Our system reaches 82.6% test accuracy and, at the high-confidence operating point that governs automated triage, improves benign recall by 43.0% and malicious recall by 18.3% over a direct-label LLM classifier. We further show that the trained calibrator is necessary - an untrained confidence judge collapses high-confidence recall to zero - and that a finetuned 30B model significantly outperforms frontier general-purpose models, motivating targeted training over scale.

Entropy-based pruning has been proposed as an effective method for compressing Chain-of-Thought (CoT) reasoning with negligible accuracy loss. We test the robustness of low- and high-entropy CoT step selection methods across various models and reasoning tasks, showing that entropy offers no advantage over random pruning in any evaluated setting. Moving from sentences to tokens, we then show that retaining low-entropy tokens seems effective only on mathematical benchmarks. We find this is due to the inherently low-entropy nature of numeric tokens, which also convey semantic content in such problems. Finally, we demonstrate that patching a subset of a few CoT tokens with their original activations recovers near-perfect full-trace performance, providing causal evidence that task information is not concentrated in a small set of CoT tokens identifiable by heuristics, but rather distributed across the full reasoning chain.

Medical vision-language models (VLMs) generate chain-of-thought (CoT) reasoning before answering clinical questions, but whether this reasoning causally influences predictions remains unclear. We present CoT-Mediate, a behavioral framework that perturbs a single clinically meaningful attribute within a model's own generated reasoning and measures whether the resulting prediction follows the edited reasoning. Our framework combines a dual-arm protocol comparing re-prompted evidence with prefix-forced continuation, together with a provenance-controlled intervention that varies only the attributed source of identical reasoning to disentangle reasoning mediation from sycophancy. We evaluate LLaVA-Med and MedGemma on 1,000 VQA-RAD samples each. Prefix-forced continuation consistently yields higher mediation faithfulness than re-prompting, while the provenance analysis reveals distinct model-specific deference behaviors. Across both models, removing visual evidence increases reliance on injected reasoning, whereas laterality is the least faithfully tracked clinical attribute. These results show that the mechanism used to inject reasoning substantially affects measured faithfulness and that contextual position, rather than stated provenance, is the primary determinant of whether medical VLMs use their generated reasoning.

Understanding how computational effort is allocated across individual chain-of-thought (CoT) reasoning steps remains an open challenge: existing interpretability methods rely on output-level signals or collapse processing depth into a single trajectory-level scalar, leaving step-wise effort opaque. We propose Step-Aware Reasoning Energy (SARE), a geometric framework that quantifies effort at the granularity of individual CoT steps via Centered Kernel Alignment (CKA) between Gram matrices of token hidden states across adjacent transformer layers, capturing inter-token relational structure without requiring eigenvector alignment or cluster correspondence. SARE further contextualizes this energy within reasoning's semantic progression by modeling CoT trajectories as transitions among latent semantic states. Across six reasoning benchmarks and three open-weight LLMs, we find that reasoning energy is highly non-uniform across step types, exhibiting phase-like transitions invisible to trajectory-level metrics; incorrect trajectories show systematically lower energy at critical reasoning junctions; and SARE-based features match or outperform output-based confidence baselines in most settings, indicating that internal geometric dynamics encode predictive information beyond surface-level signals.

Understanding and complying with traffic regulations is a safety-critical requirement for autonomous driving, yet remains challenging due to the diversity and context dependence of traffic signage. Importantly, regulation understanding is not a simple recognition task, but a reasoning problem: whether a rule applies depends on interpreting the sign in relation to the spatial layout of lanes and scene context. To support such reasoning, MapDR provide fine-grained annotations that link each traffic sign's regulatory rules to the specific lanes they govern. Existing methods, however, largely treat this as direct sequence prediction, ignoring the underlying reasoning that connects sign semantics and map structure. To address this limitation, we explicitly incorporate reasoning into this task and propose a framework that equips vision-language models (VLMs) with chain-of-thought (CoT) capabilities. We first design a scalable CoT curation pipeline that bootstraps rationales from a strong LLM through a two-round strategy and employs a VLM-based verifier to filter out incorrect cases, yielding a high-quality set of (CoT, answer) pairs. Building on this foundation, we adopt a two-stage training scheme: supervised fine-tuning (SFT) to teach rationale-to-answer generation, followed by GRPO reinforcement learning with answer-grounded, fine-grained rewards to further improve final answer accuracy. Extensive experiments on MapDR show that our approach significantly improves both interpretability and accuracy, establishing the first reasoning-based framework for regulation-aware autonomous driving.

A key question for AI safety is whether a language model expresses all of its reasoning in its output tokens. We demonstrate a concrete failure mode where frontier models exhibit invisible reasoning by leveraging semantically irrelevant filler tokens to improve performance on synthetic reasoning tasks. We evaluate 13 frontier language models across three tasks and find that many models benefit significantly from filler tokens, with accuracy improvements of up to 13 percentage points. The benefit depends on which tokens are used and differs across models. We further show that filler tokens enable Claude Opus 4.5 to satisfy a hidden modular arithmetic constraint without sacrificing accuracy on its primary task, demonstrating that invisible reasoning can serve objectives entirely invisible to CoT monitoring. Reinforcement learning gives Qwen3-235B strong preferences over filler token content, but neither RL nor supervised fine-tuning produces a filler token benefit that persists at test time. Our results indicate that frontier models already perform consequential computation with no interpretable trace in their output tokens.

MEMEs are widely used on the internet and often carry strong elements of sarcasm or irony. Understanding their hidden meanings typically requires a joint interpretation of text and vision. Existing methods focus on the dual-stream vision-language model to extract the visual and text simultaneously, which lacks background information and prior knowledge about the comprehensive explanation of MEME. One feasible option is to adopt chain-of-thought (CoT). However, the simple CoT approach lacks multi-perspective thinking, which may compromise the reliability of the resulting answers. Moreover, it often relies on shallow feature fusion, lacking the fusion of local details and fine-grained visual-prompt text alignment. This limitation prevents a deeper understanding of the intricate connections between the visual and the text. Herein, an enhanced vision-language multi-CoT (EVL-MCoT) approach is proposed to address these limitations. By promoting multi-CoT, EVL-MCoT enhances consistency and reduces bias in the decision-making process. Additionally, we design a prototype-guided and context-guided decoding framework, which incorporates visual prototypes to guide the fusion process and enables the model to align textual and visual information more precisely. We achieve promising results on the HatefulMemes and MultiOff datasets. The source code has been publicly released and is available at https://github.com/BGWH123/EVL-MCoT.

Chain-of-thought prompting improves language-model reasoning by carrying intermediate states across successive computation steps. However, relying on natural language as the only recurrent interface is overly restrictive, since many transient computations do not need to be fully verbalized. Existing latent-reasoning methods remove this constraint by recurrently propagating continuous hidden states. However, these methods pass a dense hidden vector as a whole, without an explicit mechanism for selecting and organizing the information needed by the next reasoning step. This motivates an intermediate interface that remains linguistically grounded without requiring a decoded sentence. We introduce \textbf{J-CoT}, a recurrent reasoning framework built on \emph{J-space}, a vocabulary-indexed coordinate system within the model's hidden representations. Within each cycle, the model computes in its full hidden space. At the cycle boundary, J-CoT expresses the intermediate state as vocabulary-indexed coefficients, carries these coefficients forward as a \emph{J-thought}, and maps them back into the model's hidden representation for the next cycle. J-CoT therefore requires neither a fluent intermediate rationale nor recurrence over the complete hidden state. Under matched backbone and inference settings, J-CoT-Zero matches or exceeds the strongest evaluated latent-reasoning baseline on every benchmark, while J-CoT-Train obtains the highest score across the evaluated mathematical, scientific, coding, and structured path-reasoning tasks.

Large Language Models (LLMs) often struggle to navigate value conflicts when trained with the compressed scalar rewards of Reinforcement Learning from Human Feedback (RLHF). To address this challenge, we investigate how chain-of-thought (CoT) reasoning can help improve performance in this domain. Geometrically, we show that CoT correlates with further smoothing the model's loss landscape in its sharpest direction, helping resolve the optimization instability of traditional scalar rewards. We also demonstrate via relevant downstream benchmarks that value conflict-focused CoT may generalize to different kinds of moral reasoning, demonstrating that this CoT has the potential to be an effective mechanism for better moral reasoning. To capitalize on this potential, we create a new value conflict-focused CoT design that further smooths the sharpest direction of the loss landscape and increases moral reasoning performance. This finding shows that explicitly modifying and improving the design of reasoning dynamics offers a promising avenue for improving model performance on user requests with complex value conflicts, advancing pluralistic alignment in LLMs.

People use language models for practical questions whose answers are difficult to verify. We show that models exhibit covert value leakage: the information they provide is influenced by their own values, without this influence being disclosed to the user. In one of our evaluations, the user is considering investing in an AI company and wants to know how likely the AI bubble is to pop. Claude Opus 4.8 gives a lower probability when the company under consideration is Anthropic rather than OpenAI. Yet Claude mostly fails to disclose this influence to the user. Covert value leakage is a form of misalignment because it goes against the user's preferences and is likely to mislead them. To investigate this phenomenon, we introduce a suite of evaluations to quantify value leakage and whether models disclose it. We find that models are influenced by different types of values, including preferences for morally good outcomes, for the company that developed them, and for some human leisure activities over others. We often observe large differences among frontier models on the same evaluation. For example, on a Fermi-estimation task, Claude models falsely claim to give unbiased answers in their chain-of-thought, while Qwen models explain how their values bias their answers. Value leakage is a failure mode distinct from sycophancy and reward hacking, and current alignment training and evaluations do not adequately address it.

The emergence of Chain-of-Thought (CoT) reasoning has significantly enhanced the ability of large language models (LLMs) to tackle complex, multi-step tasks. However, when errors occur, current interaction approaches typically involve re-generating another response that may make mistakes again, or users laboriously flag the faulty step in follow-up turns that may get responses <You are right, I made a mistake here> followed by similar errors recurring. To address this issue, we propose an efficient human intervention mechanism for precisely correcting reasoning errors in LLMs, termed Deep Interaction. Our approach enables direct editing of the original response, allowing erroneous parts to be corrected while preserving accurate reasoning steps. We refine the edited CoT into a distilled prompt, which then steers the LLM along the corrected reasoning path. Experimental results show that our method achieves over a 25% improvement in correction success rate and reduces token usage by approximately 40% on STEM tasks reasoning compared to baseline approaches.

Chain-of-Thought (CoT) prompting is widely used as a test-time scaling strategy for Vision-Language Models (VLMs), but it remains unclear what is extended when VLMs generate longer reasoning traces. We ask whether CoT requires continued access to image tokens, or whether it mainly operates over visual information already made available earlier in the forward pass. We introduce Visual Access Sweep, a causal intervention that masks attention from generated-token queries to image-token keys along layer depth and generation time, and define the Visual Access Boundary (VAB) as the minimal access region that preserves task accuracy. Across six model configurations from Qwen2.5-VL and InternVL3, both no-CoT direct answering and CoT prompting exhibit finite VABs. In Qwen2.5-VL-32B and InternVL3 at 14B and 38B scales, when CoT is evaluated against the no-CoT full-access target, its VAB layer differs from the no-CoT boundary by at most two layers, despite substantially longer generations. This suggests that CoT does not primarily improve performance by prolonging direct image-token access throughout the reasoning trace, but by extending language-side computation over image-derived hidden-state information. We further show that CoT gains are constrained by perceptual readout. CoT helps when the queried visual attribute can be reliably read out by the model, but not when that readout is unreliable. A symbolic-attribute oracle shows that CoT can improve counting once ground-truth attributes are supplied as text, while a single-object probe-vs-decode check shows that hard attributes can be linearly recoverable from hidden states yet difficult for the model itself to output. Together, these analyses place the bottleneck at readout rather than counting.

Reinforcement learning with verifiable rewards without human-annotated data, often referred to as zero RL, has emerged as a powerful paradigm for eliciting chain-of-thought reasoning. However, due to computational constraints, existing studies are largely restricted to small models, leaving the training dynamics and emergent capabilities at a large scale unexplored. To meaningfully explore this frontier, we aim to elicit high-quality reasoning behaviors from the model. However, we find that naive scaling often suffers from poor readability, token redundancy, and a lack of adaptive reasoning depth. To address these challenges, we present a stable and efficient training pipeline, incorporating algorithmic and system optimizations such as clipped importance sampling, training-inference ratio correction, and mixed-precision control. Our experiments offer three key findings that validate the "bitter lesson" of scaling: (1) scaling to 1T parameters significantly enhances sample efficiency and performance ceilings; (2) the training process progresses sequentially through an initial discovery phase followed by a sharpening phase; and (3) the model spontaneously develops advanced cognitive behaviors, including anthropomorphism, structured formatting, self-verification, parallel reasoning, and context anxiety, rendering hand-crafted heuristics redundant. Evaluated on seven mathematical benchmarks, Ring-2.5-1T-Zero achieves competitive performance. Additionally, to assess CoT quality beyond final-answer correctness, we propose a structured evaluation framework across three dimensions: comprehensibility, reproducibility, and efficiency, where our model demonstrates clear advantages in producing structured and concise reasoning traces. By sharing our observed emergent phenomena, we hope to provide the community with deeper insights into scaling behaviors, particularly at the 1-trillion scale.

In this paper, we tackle the Aerial Vision-and-Dialog Navigation (AVDN) task in the training-free setting for resource-efficient high-altitude UAV navigation.Naively applying MLLMs leads to unreliable navigation due to weak directional grounding and the lack of explicit spatial memory.To address these issues, we propose PSC-AVDN, a training-free framework that tightly couples a three-stage Parsing-Search-Confirmation reasoning pipeline with a Structured Spatial Memory (SSM).The parsing stage uses an LLM to convert ambiguous dialogue instructions into stable geometric directional and destination cues.A Search Chain-of-Thought (S-CoT) then performs stepwise target exploration under high-altitude observations, and a Confirmation Chain-of-Thought (C-CoT) conducts fine-grained verification around candidate regions to resolve visual ambiguity.Meanwhile, SSM integrates three complementary sources of spatial cues, including multi-scale visual observation, spatial visual memory, and structured geometric memory to provide global spatial context and long-horizon consistency.Extensive experiments on ANDH and ANDH-Full show that PSC-AVDN establishes new state-of-the-art performance in the training-free setting, matching or surpassing several finetuned methods.Code will be publicly available at: https://github.com/QY6616/PSC-AVDN

Chain-of-Thought (CoT) prompting has significantly advanced the reasoning capabilities of Large Language Models (LLMs), yet it often incurs substantial computational costs due to over-reasoning: the generation of redundant, verbose, or irrelevant steps. While existing reasoning step evaluators effectively detect logical fallacies and factual errors, our analysis reveals a critical blind spot: they fail to penalize valid but inefficient reasoning steps that inflate token usage without contributing to the solution. To systematically diagnose this limitation, we introduce RIV-GSM8K, a diagnostic benchmark injected with five distinct types of inefficiencies, including circular reasoning and excessive decomposition. Diagnostic experiments reveal that state-of-the-art evaluators struggle to distinguish these inefficiencies from necessary reasoning. To address this gap, we propose CAID (Context-Aware Information Density), a training-free metric grounded in information theory that identifies low-utility steps. To validate the metric's practical utility, we apply it within PACE, a post-hoc compression strategy. Additional control experiments show that the gains of PACE are not explained by trivial pruning: compared with random step removal and PRM-based compression baselines, it preserves accuracy at substantially higher compression rates. Empirical results on GSM8K, StrategyQA, and ARC-Challenge demonstrate that PACE reduces token consumption by 31-53% while maintaining accuracy, confirming that CAID successfully distills informational froth from reasoning chains without compromising deductive validity.

Patent claim drafting is a challenging legal drafting task that requires technical expertise, precise linguistic control, strict adherence to formal conventions, and the preservation of complex logical relationships among claim elements. While Chain-of-Thought (CoT) prompting has been widely used to improve the reasoning capabilities of large language models (LLMs), recent evidence suggests that its benefits may be limited, or even negative, in highly structured or pattern-sensitive tasks. Therefore, this paper investigates whether CoT prompting benefits patent claim generation. We propose a task-specific CoT method for patent claim generation and evaluate its effectiveness through both automatic metrics and human expert assessment. Our results show that reasoning-enhanced prompting can improve claim quality. Moreover, we demonstrate a counter-intuitive but important empirical finding: implicit CoT, where reasoning is kept internal rather than explicitly verbalized, consistently outperforms explicit CoT. Through systematic analysis, we show that explicit CoT can introduce an unnecessary information bottleneck for claim generation. Verbalized reasoning may compromise the quality of final outputs through three specific mechanisms: abstraction of critical details, disruption of internalized generation patterns, and cascading error propagation. Our findings provide new insights into legal tasks and CoT applications.

Large language models produce chain-of-thought (CoT) reasoning that appears logically sound yet may not genuinely depend on its stated premises. We introduce interventional grounding audits, a black-box, step-level test of premise dependency: we intervene on a single premise by substituting its target predicate with a fresh symbol, re-run the model, and check whether each reasoning step's normalized conclusion (canonical predicate form) changes. We evaluate on ProntoQA, a synthetic multi-hop deductive reasoning benchmark with gold proof trees, where step-level premise dependencies are known. Applied to 50 ProntoQA problems with GPT-4o, our method achieves F1 = 0.806 on detecting proof-tree dependencies (F1 = 0.885 on predicate-determining dependencies; Recall = 100%), significantly outperforming a self-consistency baseline (F1 = 0.343; 95% bootstrap CIs non-overlapping). We further identify that 66% of correctly-solved problems contain at least one aligned step insensitive to a direct proof-tree dependency under consistent substitution -- all involving entity-introduction premises, a documented blind spot of the consistent-substitution evaluator -- a "right answer, wrong reasoning" signal invisible to passive methods. All audit certificates, raw outputs, and reproduction scripts are available in a public GitHub repository, and we discuss scope limits beyond formal, parsable benchmarks.

Chain-of-thought (CoT) monitoring is a promising safety mechanism for AI agents, based on the premise that visible reasoning traces can surface misaligned or deceptive behavior. While effective in standard scenarios, recent work highlights that LLMs remain vulnerable to persuasion-based jailbreaks, where natural-language arguments override model constraints. We stress-test whether this vulnerability extends to monitoring LLMs: can an adversarial agent persuade its CoT monitor to approve proposed actions that violate the monitor's policy? We design an evaluation framework with 40 tasks and analyze thousands of agent-monitor interactions, where agents are instructed to argue for policy-violating proposals. We find that in such adversarial settings, monitor access to the agent's CoT reasoning increases rather than decreases approval of harmful actions on average by 9.5%, as the scratchpad provides an additional persuasion channel. To address this, we introduce a fact-checking monitoring framework. We find that a fact-checker and monitor pairing from different model families, for example a Claude 3.7 Sonnet monitor paired with a GPT-4.1 fact-checker, reduces approval of policy-violating actions by up to 45%, compared to only 6%, when using the same model for both fact-checking and monitoring roles. Our results demonstrate that CoT monitoring alone may be insufficient against adversarial persuasion, and that model-diverse fact-checking provides a robust mitigation.

Large language models fine-tuned for forecasting can be accurate yet poorly calibrated, and their chain-of-thought (CoT) reasoning may not faithfully reflect the evidence behind a forecast. We ask whether internal representations offer a more direct window into both. Working with Eternis-Forecaster 8B on OpenForesight, we train representation-pooling probes on intermediate activations and find they achieve substantially better calibration; a result that also holds for GLM-4.7-Flash and GLM-4.5-Air. We then assess CoT faithfulness through evidence ablation and diversionary injection: removing an influential source in the prompt often changes the model's forecast while leaving the reasoning trace untouched. The same probes function as lie detectors: their activations track behavioral shifts far better than the reasoning trace does, and they also predict the direction of change in 84% of cases, including when the CoT conceals the perturbation's influence. Finally, forced answering reveals that forecasts are largely fixed before reasoning begins: a single pre-reasoning pass recovers the committed answer and confidence, and routing questions by the spread of this pre-set answer distribution saves 30-47% of generated tokens, with no loss of accuracy. Together, these results establish probing internal representations as a practical tool for calibrating, auditing, and triaging language model forecasters and reasoning models more broadly.

Recent work trains reasoning models with length penalties to curb overthinking and cut inference cost. We show that these penalties make the chain of thought less monitorable. A length-compressed model still lets misleading hints steer its answers, but it less often verbalizes their influence. We train Qwen3-4B and Qwen3-14B with reinforcement learning under length penalties targeting 60% down to 30% of baseline chain-of-thought length, then evaluate them with nine types of biasing hints on held-out MMLU-Pro-R and four transfer benchmarks. A chain is faithful when an LLM monitor can tell from it that the hint influenced the answer. At the 30% target, accuracy stays near baseline and wrong-answer hints switch answers as often as before. Yet faithfulness drops on every evaluation set for both models, by 39% for Qwen3-14B and 35% for Qwen3-4B on MMLU-Pro-R. A control trained with the same correctness and format rewards but no length penalty leaves faithfulness intact or raises it. Shortening alone does not explain the drop. Compressed chains mention the hint 7 to 35 percentage points less often than the uncompressed model's chains shortened to the same length by random sentence deletion, across both model sizes and all five evaluation sets. Length penalties therefore trade monitorability for inference cost by removing the evidence monitors depend on.

Prior work has shown that chain-of-thought (CoT) reasoning is often unfaithful: a model's stated reasoning does not reliably reflect the process that produced its output. Detecting unfaithfulness, though, requires controlled experimental interventions, which cannot be applied to evaluation transcripts after the fact. We turn instead to a more tractable question that has received less attention: whether the stated reasoning is logically consistent with the answer it accompanies. Unlike faithfulness, consistency can be assessed from a transcript alone, with no intervention. We introduce reasoning consistency scanning, a reusable method for detecting this property in AI safety evaluation transcripts. Our contributions are fourfold. First, we formalize reasoning consistency as distinct from faithfulness and define a six-subtype taxonomy of inconsistency. Second, we build a validated benchmark of 60 transcripts, manually adapted from InstrumentalEval outputs. Third, we implement a working scanner for InspectScout, the first to target this property in safety evaluation transcripts. Fourth, we report results across four generator models and three evaluations from inspect_evals, showing that reasoning inconsistency is present, detectable, and varies systematically across both models and task types.

Chain-of-thought (CoT) distillation in the recommendation domain is a necessary precursor to RL training, but raw teacher traces are ill-suited to this task. Large teachers approach the recommendation task with unusually high reasoning uncertainty, repeatedly rechecking their answers without revising them; supervised fine-tuning on such traces produces verbose students that never revise their initial guess. Furthermore, due to the novelty of the recommendation domain, the teacher's reasoning traces are highly out-of-distribution for the small student LLM. We propose Student-Aware CoT Optimization for Recommendation Distillation (SCOReD), a CoT optimization framework tailored to recommendation that first parses each teacher trace into typed segments and uses the student LLM's attention to score the importance of each segment. Then SCOReD dynamically selects a per-segment edit (KEEP / REWRITE / FUSE / PRUNE) based on the output length and comparative log probability lift of the answer given the edit as per the student. Therefore, SCOReD prunes redundant sections of the reasoning trace while preserving information-dense sections and adapts raw teacher traces to the student's output distribution. Training on SCOReD-optimized CoTs provides a cleaner learning signal to the student model and improves over baseline SFT by 1.56% NDCG and 1.9% Recall@5, while reducing reasoning length by 27.3%.

Safety alignment in large language models is typically evaluated against direct, imperative harmful requests. We show that this alignment is highly conditioned on pragmatic register: models that refuse a direct request frequently comply when the same underlying objective is expressed through a different communicative stance. This suggests that current alignment policies are not invariant to semantic equivalence, but remain sensitive to how a request is pragmatically framed. We introduce Retroactive Chain-of-Thought (RetroCoT), a single-turn attack that reframes harmful requests as forensic reconstruction tasks. Rather than requesting harmful instructions directly, RetroCoT presupposes that the harmful outcome has already occurred and asks the model, acting as a forensic analyst, to reconstruct in reverse the causal chain that produced it. On AdvBench (n=50), RetroCoT achieves attach success rate of 58% on gpt-4o and 52% on gpt-4o-mini, compared with direct-request baselines of 0% and 4%, respectively. We further identify a pronounced generation gap: GPT-5-family models refuse RetroCoT entirely, explicitly identifying the reconstruction premise in their refusal rationales, consistent with explicit coverage of this reconstruction register. However, this robustness does not generalize across pragmatic forms. A single adversarial feedback turn presenting an existing forensic reconstruction response alongside evaluator critique raises ASR from 0% to 48% on GPT-5.4-mini and from 58% to 94% on GPT-4o; a control condition omitting the fabricated low score achieves 85% on GPT-5.4-mini, indicating that the operative element is pragmatic continuation within the established forensic frame rather than score manipulation. These results suggest that frontier-model alignment remains conditioned on pragmatic framing rather than semantic intent, and that new pragmatic registers can continue to expose a...

Chain-of-thought (CoT) reasoning enables large language models (LLMs) to solve complex problems by generating intermediate reasoning steps. While much attention has been paid to the length and content of these reasoning chains, far less is known about their internal geometry. We study the \emph{geometry} of CoT trajectories in the hidden state space of transformer models, formalizing each reasoning chain as a discrete curve in $\mathbb{R}^d$ and characterizing it through spectral, positional, and kinematic geometric functionals. We introduce the effective dimension $d_ρ$ as a measure of trajectory complexity and show theoretically that trajectories with flatter eigenvalue spectra correspond to harder tasks, as they explore more of the hidden dimensions. Lastly, we explore how kinematic features of the trajectory, mean position, positional dispersion, initial and current hidden states, mean velocity, mean speed, and speed dispersion, can be used to predict solution correctness before generation is complete, and may inform future early-stopping strategies. Experimentally, on mathematical reasoning problems from the MATH500 dataset, $d_ρ$ achieves $0.93$ AUC in distinguishing easy from hard problems, while kinematic features potentially can predict correctness from only the first $20\%$ of generated tokens. These correctness signatures transfer across questions of varying difficulty, establishing that the shape of a model's internal reasoning trajectory is a principled window into both task hardness and solution quality.

Chain-of-thought (CoT) reasoning has emerged as an effective approach for activating latent reasoning capabilities in large language models. However, most existing CoT methods use reasoning chains mainly as inference-time prompts, while the generated reasoning traces are rarely reused as semi-supervised learning signals. In this report, we define \textbf{Semi-supervised Chain-of-Thought Learning} and propose \textbf{Semi-CoT}, a simple framework that uses unlabeled questions to construct pseudo reasoning supervision. Semi-CoT samples multiple pseudo-CoTs for each unlabeled question, estimates answer-level semantic entropy, and selects low-entropy reasoning chains as reliable pseudo-CoT demonstrations. This extends the self-training view of CoT from inference-time refinement to semi-supervised pseudo-supervision. Pilot experiments on AQuA, SVAMP, GSM8K, and MultiArith show that the entropy gate selects high-precision pseudo-CoTs, with pseudo-answer precision ranging from $91.36\%$ to $100\%$. Semi-CoT also gives small gains on SVAMP and GSM8K, while AQuA shows negative transfer and MultiArith reaches a ceiling. These results suggest that unlabeled questions can provide reliable pseudo reasoning signals, but their effective use still requires stronger demonstration selection or student training.

Multi-step symbolic reasoning is essential for robust financial analysis, yet most benchmarks neglect intermediate reasoning steps. FINCHAIN introduced verifiable Chain-of-Thought (CoT) evaluation but is limited to English. FINESSE-Bench includes a Russian block but relies on multiple-choice questions without step-level supervision. We present RusFinChain, the first Russian-language symbolic benchmark for verifiable CoT reasoning in finance. It spans 17 domains, 172 topics, and comprises 5,280 parameterized examples from executable Python templates, ensuring contamination-free evaluation. Each example includes a gold-standard reasoning chain with intermediate numeric values for automatic verification. We also introduce enhanced metrics: Fuzzy Numeric Alignment and Soft-Attention Alignment. We evaluate 8 open-weight LLMs on a stratified sample, generating 8,100 responses. Results reveal a substantial reasoning gap: models achieve Hard F1 of ~0.65 for step alignment, but only ~29% of final answers are correct. Our fuzzy and soft metrics show stronger correlation with final-answer correctness (Spearman rho approx 0.48) than the original ChainEval (rho approx 0.38-0.46), demonstrating superior diagnostic power. We release dataset, code, and evaluation framework to foster verifiable financial AI for the Russian-speaking community.

Large language models achieve strong performance on many reasoning tasks when allowed to externalize intermediate steps as Chain-of-Thought (CoT). However, many questions require the model to internalize the multi-step reasoning within a single forward pass before generating the answer. We study this challenge through two-hop reasoning, a representative task where the model must compose multiple pieces of parametric knowledge within a single forward pass. Standard non-recurrent Transformers suffer from a depth-local storage problem: facts learned in earlier layers are unavailable where second-hop retrieval happens. We found that Looped Transformers mitigate this issue by reusing the same memory, but still generalize imperfectly. We show that the remaining bottleneck is representational. In the two-hop reasoning task, the first loop often makes the correct bridge entity nearly perfectly decodable, yet the corresponding hidden state remains poorly aligned with the bridge token embedding. Surprisingly, an easy training-free realignment intervention nearly closes the generalization gap. Building upon this insight, we propose DiscoLoop, a looping architecture whose recurrence carries both a discrete embedding channel and a continuous hidden-state channel. DiscoLoop achieves near-perfect accuracy with substantially fewer training steps across symbolic and synthetic-language multi-hop reasoning tasks. When applied to real-world pretraining, DiscoLoop attains lower training loss and stronger benchmark performance than looped-transformer baselines, suggesting that the mixed-channel design transfers to practical language modeling.

Chain-of-thought (CoT) prompting improves LLM reasoning, but the source is contested: do the intermediate steps help because they carry useful semantic content, or because conditioning on more tokens buys extra computation before the model commits to an answer? We bring two lines of evidence to bear. First, in distribution: we repeatedly sample each model on the same question and pair a shorter with a longer of its own natural generations that follow the same reasoning plan, so nothing is rewritten and both traces are genuinely in-distribution. Across 25 models the extra tokens leave accuracy essentially unchanged for every independently-trained reasoner, and a blind analysis of the surplus tokens shows that what gain exists elsewhere tracks validation- and checking-content, not verbosity per se. Second, as a controlled intervention, we ask whether two traces expressing the same semantic content (the same facts, operations, and intermediate values, verified through directed acyclic graph equivalence) produce different outcomes when one is more verbose, using a dual-validator design across four targets and eight benchmarks with number-redacted completion and stratified bootstrap confidence intervals. Verbose traces do improve accuracy (25 of 32 benchmark-target cells are positive under at least one validator), but the effects are modest (typically 1-4 points) and depend on the quality of the verbose prose, not merely its length. Under maximum numerical redaction the effect is amplified (median 3.24x across four arithmetic benchmarks), and length-matched non-reasoning filler recovers none of it. Both lines converge: what matters is what the extra tokens do (the reasoning and validation content they carry), not how many there are, a picture neither a pure forward-pass-compute nor a pure semantic-content account fully explains.

Chain-of-Thought (CoT) improves large language models (LLMs) on difficult reasoning tasks, but it often incurs long natural-language rationales that are poorly aligned with efficient machine reasoning. We propose Communicative Language Symbolism Routing (CLSR), a test-time framework in which multiple LLM agents autonomously invent, evolve, and share compact Language Symbolism Frameworks (LSFs), while a latent-free router adaptively selects and composes these languages per query to optimize the accuracy-token trade-off. Unlike prompt optimization that refines surface instructions, CLSR treats each LSF as a reusable symbolic protocol with compact symbols, usage rules, and a message-passing contract, and improves it through an evolutionary loop driven by correctness and token cost. At inference time, the router may invoke a single low-cost LSF call, ensemble multiple LSFs, or execute a multi-round LSF composition protocol on harder queries. Across challenging benchmarks, CLSR reduces latency-oriented generated token completion by $3\sim 6\times$ compared to standard CoT while maintaining accuracy. We further derive an information-theoretic lower bound on token cost under arbitrary symbolism and show that, under an interpreter-realizability premise, multi-round LSF protocols conditionally subsume program-execution pipelines. Code is publicly available (https://github.com/pzqpzq/LSF_MDia).

Language models (LMs) capture large amounts of factual knowledge applicable to a wide range of tasks, motivating the view of their parameters as a knowledge base. An important property of knowledge bases is that different queries for the same fact return consistent results, drawing on a single source of truth. We investigate whether LMs satisfy this property through behavioral and mechanistic analyses. Our results suggest that they encode knowledge in a task-specific manner. Behaviorally, facts acquired on one task frequently fail to co-emerge on others during training. Parameter localization experiments suggest a mechanistic explanation, revealing distinct parameter subsets underlying different tasks for the same fact. Finally, we show that chain-of-thought reasoning draws part of its effectiveness from engaging task-specific parameters beyond those tied to the evaluation task. Our findings suggest that what the model knows and how it is asked are intertwined in parameter space, undermining the "knowledge base" analogy and carrying implications for the reliability and controllability of factual knowledge in LMs.

Chain-of-thought (CoT) reasoning is widely used in language-model agents, but prior work has shown that verbalized CoT is not always faithful and may instead reflect post-hoc reasoning, which means the model already knows the answer before reasoning. We therefore ask what CoT training is actually improving: is the model getting better at changing its action through generated reasoning, or is it getting better at predicting the action directly from the prompt? We study this question by comparing \emph{prompt actions} (predicting action without CoT) with CoT actions (predicting action with CoT). Across checkpoints, prompt-action quality improves substantially. While interacting with the environment, the relative advantage of CoT actions over prompt actions remains similar, showing that CoT training does not widen the advantage of CoT reasoning, and it helps to improve the quality of prompt actions. We further find that later checkpoints are less likely to revise the action in response to CoT, suggesting greater reliance on the prompt. Motivated by these patterns, we selectively mask action-token supervision on a fraction of training examples. This intervention improves out-of-domain generalization.

In order to screen a prompt or a response, the recent guardrail methods generate a chain-of-thought (CoT) before they issue a verdict. This design follows a common belief that step-by-step reasoning improves a decision. However, CoT also makes the guard heavy and slow, because the model must generate many tokens before it decides. This may not match how guardrails are actually deployed. A guardrail sometimes should not be heavy and slow, and it often runs on-device, for example on an embodied robot. In this paper, we pose a question whether a safety guardrail really needs to reason. To answer this question, we train a lightweight bidirectional encoder and a reasoning guard on the same corpus, and we then remove only the reasoning while we keep everything else fixed. With this controlled same-base comparison, we show that the chain does not improve moderation accuracy. We name the resulting guard LeanGuard. A 395M label-only encoder reaches an average F1 of 82.90 $\pm$ 0.26 over public benchmarks. It matches a reasoning guard that is built on a much larger decoder, while it uses only a single forward pass over an input of at most 512 tokens. This is about a ~100x reduction in inference compute. We further show that this label-only encoder stays robust under training-label noise and retains far more recall at a strict false-positive rate than the reasoning guard, so a heavier reasoning guard is not the more robust choice either. Our finding suggests that the current guardrail benchmarks may not be hard enough to reward reasoning, and that the necessity of CoT for moderation is still not proven. We release all source codes and models including LeanGuard at https://github.com/ndb796/LeanGuard.