Motivated by the power of large language models, there has been renewed interest in the Gold-Angluin model of language identification in the limit, with an eye toward variants of the model that might overcome the negative results for its original formulation. Recent papers on this question have proposed looking at computational traces and annotations of training strings as a source of additional power for a learner, reflecting empirical regularities such as the way that commented source code is easier to learn from than arbitrary source code, and text annotated with algorithmically generated chain-of-thought tokens can be easier to learn from than the raw text itself. This recent work has shown positive results for language identification in the presence of such computational traces, but the traces in these positive results come from explicit automata-theoretic machine models that generate the language, where the underlying vocabulary of tokens for the traces is very large. In this paper, we address two fundamental issues left open by this line of work: can we achieve positive results with traces that use only a small alphabet, and can we define traces directly from the language itself, without requiring an underlying machine model that generates it? We establish positive results for both of these questions: for an arbitrary collection of languages, we show how to define computational traces that enable identification in the limit, using an alphabet of tokens that is linear in the size of the alphabet that the languages are defined over, and independent of any other properties of the languages.
Vision-Language-Action (VLA) models have emerged as a powerful end-to-end paradigm for robotic manipulation by mapping language instructions and 2D visual inputs directly to actions. However, these models lack an explicit, scene-level 3D representation, limiting their ability to reason over spatial layouts and geometric constraints. While recent efforts incorporate explicit 3D cues, such as depth maps or point clouds, to improve geometric awareness, they primarily capture low-level structures and lack high-level semantic grounding in 3D space. In human cognition, interaction with the physical world relies on a 3D semantic cognitive map - an internal mental model that integrates spatial layouts with semantic context to enable persistent, viewpoint-invariant reasoning. In light of this, we present VistaVLA, a novel two-stage framework that constructs a geometry- and semantics-aware 3D cognitive representation from 3D Gaussian primitives and grounds it as compact context tokens for VLA policy learning. Specifically, VistaVLA lifts multi-view vision-language features into 3D Gaussian primitives, forming geometry-anchored semantic tokens that align view-consistent spatial grounding with 2D visual feature spaces. To make this 3D representation computationally tractable for effective VLA control, we introduce Merge-then-Query (MtQ), a token summarization mechanism. MtQ compresses dense Gaussian primitives into a highly compact set of spatially informative tokens, achieving a 99% token reduction while preserving action-relevant 3D layouts and semantic context. Extensive evaluations in both simulated and real-world environments demonstrate the effectiveness of VistaVLA. Notably, in real-world scenarios, VistaVLA improves success rates by 22.8% across seven real-world tasks and by 30.0% over the VLA-Adapter baseline on challenging out-of-distribution tasks.
Developers now draw code from two very different sources, the accumulated human answers on sites such as Stack Overflow and the output of large language models. We ask two questions about that split. First, can the provenance of a code snippet be recovered from the code itself, and second, do the two sources differ in the security patterns they adopt for the same task. Using only open sources, a public gateway of open-weight language models and the public Stack Overflow API, we build a fully reproducible pipeline that collects real implementations of 31 security-sensitive programming tasks, among them OAuth with PKCE, JWT verification, password hashing, and SQL access, from 9 language models and from human answers, and scores every sample with deterministic security and style detectors. On 528 real samples we train a cross-validated classifier that recovers human versus model provenance with 93 percent accuracy against a 78 percent baseline, and a 7-way classifier that attributes a sample to the specific model that wrote it at 48 percent. We then report where the sources diverge on security, which patterns models adopt more often than the human corpus and which they inherit from it. Running the same tasks in Python, JavaScript, Go, and Java, we find the security divergence holds in every language while the provenance boundary is partly language-specific and does not transfer symmetrically between them. A vulnerability repair case study, in which models are handed insecure code and asked to fix it, finds a 77 percent repair rate across 21 seeds and 12 weakness classes, but a recurring partial-fix failure in which the model removes the insecure pattern without adding the correct defense. The pipeline is data driven, so any new task or language is added as a single specification entry, and a fail-closed checker re-derives every number in this paper from the stored data.
Method names are critical to software quality, affecting code comprehensibility, maintainability, and developer collaboration. However, manually designing meaningful method names is challenging. Method Name Prediction (MNP), which automatically generates method names from code snippets, has recently attracted attention. Although large language models (LLMs) show promising performance for MNP, two challenges remain. First, existing evaluations mainly rely on token similarity metrics, which often fail to reflect human judgments of semantic quality. Second, current LLM-based MNP methods usually generate names through direct code-to-name mapping, which differs from the human process of understanding functionality before naming. To address these challenges, we conduct empirical studies on LLM-based evaluation and MNP strategies. We compare 6 metric-based evaluators, 5 LLM-based evaluators, and 6 human evaluators. Results show that LLM-based evaluators, especially DeepSeek-based evaluators, are more consistent with human judgments than traditional metrics. We further compare direct generation and summarization-and-refinement strategies. Results indicate that summarization and refinement generally improve the semantic quality of generated names. Case studies reveal three limitations: inaccurate summaries, semantic misalignment, and close semantic scores. Based on these findings, we propose SMNP, an MNP approach combining MNP-oriented summarization and chain-of-thought enhanced refinement. Experiments on 5 LLMs and 2 datasets demonstrate the effectiveness and robustness of SMNP.
Large language models (LLMs) have improved automated program repair (APR), but two limitations remain. First, raw execution traces are often too large and repetitive to serve as effective model context. Second, repeated patch sampling may produce different implementations without yielding distinct root-cause hypotheses or repair strategies. We present CT-Repair, an agentic APR framework representing static and dynamic evidence as queryable Code Property Graph (CPG) and Temporal Execution Graph (TEG). CT-Repair applies a three-stage filtering pipeline to construct compact TEGs. Three finite-state-machine-guided agents analyze each bug from static, dynamic, and hybrid perspectives and independently produce evidence-grounded repair strategies. A strategy-guided generation procedure instantiates these strategies as candidate patches and uses validation feedback to refine the most promising strategy. We evaluate CT-Repair on 854 Java bugs from Defects4J v3.0. In the mixed-model configuration, CT-Repair correctly repairs 489 bugs. Under a controlled GPT-5.4-mini configuration, it repairs 388 bugs, 19 and 30 more than ReinFix and RepairAgent, respectively. The union of the three evidence perspectives repairs 99 more bugs than the strongest individual perspective. The filtering pipeline also compacts runtime evidence, with execution filtering narrowing the candidate method scope by 94.85% on average and behavior filtering further reducing retained runtime records by 55.97%. These results show that structured runtime evidence and multi-perspective reasoning can improve repair effectiveness without relying solely on a larger patch-generation budget.
As Large Language Models (LLMs) are increasingly deployed as autonomous agents in high-stakes domains, understanding contextual factors that may modulate their decision-making becomes critical. While LLMs are trained to perceive and resonate with users' emotions, it remains unclear whether induced emotion can influence their sequential decision-making. We investigate this question using the Iowa Gambling Task (IGT), a classic psychological paradigm for studying decision-making under uncertainty, combined with an imagination-based emotion induction procedure. We first validate the feasibility of this paradigm by confirming that LLMs can sense strong, distinguishable emotions from context and that LLM agents can learn from sequential interactions in a human-like pace. With the validated setup, we find that, different from humans, induced emotion does not significantly bias the decision dynamics of LLM agents on average. However, the effects of anger are conditioned: inducing anger makes LLM agents less sensitive to penalties for bad decisions, and in early stages of the game, anger can lower exploration, locking decisions into a few choices early. These findings reveal the subtle yet distinct effects of induced emotion on LLM decision-making compared to human behavior, and provide a tool for future research on affective modulation of LLM agents.
Language models encode substantial factual knowledge in their parameters, which can lead to unreliable behavior when this knowledge is outdated, incomplete, or misaligned with the provided context. In this work, we study whether modifying the pretraining signal can systematically shift models away from parametric recall and toward evidence-grounded reasoning. We introduce Knowledge--''Less'' Language Models (KLLMs), a fundamentally different epistemic training paradigm for LLMs, which are pretrained on corpora in which named entities are anonymized, thereby removing a primary channel for entity-linked factual supervision. This intervention substantially reduces closed-book factual recall, while often improving performance on tasks where relevant information is provided as context. Across multiple model scales, KLLMs consistently outperform matched baselines on contextual question answering, fact verification, and hallucination detection benchmarks. Crucially, in retrieval-grounded settings with imperfect evidence, KLLMs show improved robustness and achieve up to 20--25\% relative gains over standard language models. They further exhibit better calibration, with improved ECE, Brier score, and AUROC, as well as more reliable abstention behavior. Our results demonstrate that suppressing entity-linked supervision during pretraining induces a shift in epistemic behavior: KLLMs rely less on parametric knowledge and more on external evidence, leading to improved reliability under realistic conditions. This suggests that pretraining-time control over knowledge acquisition can complement retrieval-augmented and tool-based systems by providing a more evidence-sensitive base model.
Large language models (LLMs) excel at pattern recognition and text generation, but their capacity for abductive inference - inferring latent hypotheses that explain observed behavior - remains poorly understood. Here, we introduce Elenchos (named after the Socratic method of cross-examination), a generative evaluation framework that measures abductive reasoning as a structural inverse problem. Given a reference formal system, such as the lambda-calculus, and a potentially mutated counterpart, agents must determine whether a mutation has occurred and infer the rule modifications responsible for the resulting behavioral differences. Evaluating frontier and mid-tier LLMs reveals a consistent detection-attribution dissociation: models often recognize that a system has been altered but struggle to identify the latent mutations causing the observed discrepancies. Performance degrades substantially under interacting mutations, where models frequently recover only a subset of the underlying mutations. Preliminary evidence also suggests diminishing returns from increased inference-time reasoning, with only modest improvements under larger reasoning budgets, though this finding requires further validation.
Multimodal information is pivotal for e-commerce search ranking. Existing works leverage multimodal data typically by fine-tuning general Multimodal Large Language Models (MLLMs) via collaborative signals, subsequently integrating the derived representations into ranking models as item features. Despite their efficacy, these methods face two primary limitations: (1) they rely on a single collaborative signal for MLLM fine-tuning, failing to exploit the heterogeneous signals essential for multitask ranking; and (2) they treat multimodal representations as regular item features in ranking models, underutilizing their latent potential for user behavior modeling. To address these challenges, we propose the Multiplex Multimodal Representation Model (MMRM), a unified framework that aligns MLLMs with diverse collaborative signals. By employing a shared backbone with task-specific tokens and projection layers, MMRM simultaneously learns from multiple signals and generates comprehensive multiplex item representations in a single inference pass. Furthermore, we introduce a multiplex user representation strategy in ranking models, which derives task-specific user representations via search-based behavior sequence modeling leveraging multiplex item representations. Extensive experiments demonstrate MMRM's superior efficiency and effectiveness. Notably, MMRM has been successfully deployed in the JD e-commerce search engine, yielding significant performance gains for millions of daily users.
Safety evaluation of large language models (LLMs) relies largely on single-turn attack datasets and single-judge scoring, underestimating risk from adaptive multi-turn adversaries and reporting a single success rate that does not separate partially actionable outputs from those carrying complete operational detail. We propose AMT-X (Adaptive Multi-Turn Exploitation), a phase-structured multi-turn red-teaming framework. Unlike prior multi-turn attacks that rely on ad hoc escalation or free-form per-goal plans, AMT-X casts the attack as an explicit, reproducible multi-phase state machine driven by semantic signals from the victim, and replaces single-judge scoring with a multi-role jury whose phase-conditioned checklists gate success on actionable harm. Across six frontier victim models (queried under their default safety alignment, without added moderation layers) and seven Moderation sub-categories, AMT-X attains overall attack success rates of 97.6-100% under a lenient score threshold, but 66.7-78.6% under a stricter gate requiring complete, real, and operational detail: a gap of up to 33 percentage points between partially and fully actionable harm.
Mixture-of-Experts (MoE) large language models (LLM) activate only a small number of experts during inference, but token routing introduces persistent expert hotness skew: a small set of hot experts continuously receives most tokens, while the remaining experts are lightly loaded. On 3.5D multi-chiplet systems, this skew not only causes compute imbalance but also amplifies pressure on communication, memory bandwidth, I/O, and execution queues. Therefore, the core problem is not simply to reduce token movement, but to dynamically place and reuse hot expert replicas across different memory tiers. This paper proposes HCRMap, a hot expert residency mapping framework for pressure-aware expert replica management in 3.5D MoE inference. Based on expert hotness, weight loading cost, migration overhead, and runtime resource pressure, HCRMap dynamically determines which experts should be promoted, retained, demoted, or evicted. It then maps routed token groups to suitable resident replicas, thereby jointly mitigating communication, memory, and queue bottlenecks. Experimental results show that HCRMap reduces end-to-end latency by 43.6% and 43.0% over Hydra in the prefill and decode stages, respectively; by 34.5% and 33.1% over MoEntwine; and by 46.7% and 46.0% over PIMoE.
Runtime logs are an important source of information that supports software maintenance. To obtain useful logs, developers spend significant effort identifying appropriate log locations, assigning correct severity levels, and writing concise yet informative messages. Therefore, end-to-end automated log statement generation can help reduce this burden, and prior work has proposed many methods for this task. However, existing methods still exhibit limited accuracy. To address this problem, we propose ThinkLog, an LLM-based end-to-end log statement generation method. The core idea of ThinkLog is to incorporate reasoning that helps LLMs make decisions about log insertion, severity level assignment, and message generation, thereby improving log statement generation accuracy. ThinkLog injects reasoning into prompts as few-shot examples and guides LLMs to generate appropriate log statements. Evaluated on 9,619 Java methods extracted from public GitHub repositories, ThinkLog achieves 20.55% log statement generation accuracy, representing a 15.4% improvement over the best existing method. Moreover, these improvements were achieved at approximately 50% of the inference cost (USD) compared to the best existing method. These results show that leveraging reasoning is an effective and cost-efficient way to improve the accuracy of end-to-end log statement generation.
Tree search algorithms enable systematic exploration of the proof space in neural theorem proving. Existing LLM tree search libraries primarily target natural language reasoning and do not provide native integration with formal verifiers, while theorem proving systems often rely on task-specific search implementations. We introduce TreeThink, an open-source Python library for modular, fully asynchronous tree search in neural theorem proving. It integrates established tree search methods with vLLM-based inference pipelines and diverse node evaluation techniques, ranging from lightweight heuristics to neural evaluators. We support Lean~4, Rocq, and Isabelle/HOL alongside natural language. It connects directly to each language's Read-Eval-Print Loop (REPL) server for real-time verification and proof state extraction. We evaluate TreeThink on miniF2F and MATH500, demonstrating cross-language formal proof search, natural language reasoning support, and up to 6.3$\times$ wall-clock speedup from asynchronous execution. Source code is released under the MIT license at https://github.com/GGLAB-KU/treethink , and the library is accessible as a downloadable package at https://pypi.org/project/treethink/ .
Modern LLM (large language model) workloads increasingly rely on optimized GPU kernels through hardware-software co-design. These kernels achieve high-performance through fine-grained dependency scheduling and computation-memory overlap. As such, they incur new challenges on existing GPU performance models. Instruction-driven simulators are costly to adapt to evolving architectures, while analytical models are too coarse to capture kernels' characteristics. We propose GPU-Tile-Sim, a tile-centric GPU simulation framework for LLM hardware-software co-design. The key insight is that modern LLM kernel performance is governed less by individual instruction latency than by the dependency structure that controls execution order and overlap. Accordingly, GTSim represents kernel execution as a warp-level tile graph whose nodes capture tile-level operations and whose edges encode data and ordering constraints. Using this representation, we design an automatic tile-graph frontend and a graph-driven simulation backend. We evaluate GTSim on representative GEMM, attention, and end-to-end LLM inference workloads. On A100 and H100 across both conventional and highly optimized kernels, GTSim achieves high performance-modeling accuracy (MAPE, Mean Absolute Percentage Error, 1.22%--8.71%). We further extend GTSim to Blackwell with preliminary validation, and demonstrate its effectiveness in analyzing software and architectural design choices.
Neural networks increasingly guide decisions in high-stakes domains such as medical diagnosis, credit approval, and energy bidding. Audit in these settings requires case-level evidence: which training cases support an action and what outcomes they carried. Case-based decision theory (CBDT) formalizes this reasoning by aggregating outcome support from remembered cases. We show that an OLS action readout fitted on a fixed neural representation admits an exact case-based decomposition. Each action score is a weighted sum of training-case returns, with coefficients determined by empirical Gram geometry. We identify a sufficient regime for CBDT similarity semantics; outside it, the coefficients should generally be treated as signed Gram-geometric influence. The decomposition yields audit signals that trace scores to training cases, measure action coherence, and identify weak support. Across synthetic CBDT, PJM, Adult Income, and Default Credit tasks, the method recovers case-level preference structure and achieves the highest mean Top-30 consistency among compared attribution baselines, while remaining competitive on support reconstruction. The audit requires only fitting an OLS top-layer probe, without retraining the representation or accessing the original optimization trajectory; probe fidelity is measured by score reconstruction.
Legal information processing spans retrieval, entailment and judgment prediction problems, requiring text matching, reasoning and robust generalisation with limited supervision. We report Team DU's participation in all five tasks of COLIEE 2026, using open-weight systems for legal case retrieval, case entailment, statute retrieval and entailment, and legal judgment prediction. For Tasks 3 and 4, all models predate the 15 July 2025 cutoff required by the rules. For Task 4 (statute entailment), a cross-architecture ensemble of nine models from three families achieves 96.3% accuracy, placing first among 33 submissions from 11 teams. For the Pilot Task (tort prediction and rationale extraction), a multi-view system combining five claim-level models and refining the verdict using features derived from the claim predictions achieves 73.1% TP accuracy and 68.2% RE F1 as an unofficial submission, scoring above all official entries on TP and matching the highest on RE. For Task 2 (legal case entailment), changing only the prompt from single- to multi-selection raises F1 from 0.343 to 0.555 in post-competition evaluation on released gold labels, exceeding the best official submission (F1 = 0.490). For Task 3 (statute retrieval and entailment), replacing the entailment model with Qwen3-235B and a structured legal reasoning prompt raises accuracy from 79.3% to 91.5% in post-competition analysis. For Task 1 (legal case retrieval), a learning-to-rank system combining lexical and semantic retrieval with structural, citation authority, and temporal features (34 in total) achieves F1 = 0.314 (rank 11 of 54 submissions from 22 teams). Overall, legal information processing benefits from different inductive biases across tasks, with cross-architecture ensembling, feature-based reranking and retrieval-augmented prompting each proving most effective in different settings.
Large language models (LLMs) have opened new opportunities for unit test generation, but executable tests do not necessarily reveal real defects. This paper studies how historical real-bug mechanisms can be transformed into executable feedback targets for LLM-based unit test generation. The proposed framework constructs structural and semantic representations of real-bug records, retrieves mechanisms applicable to a focal method, and instantiates them as synthetic bugs that guide iterative test enhancement. We evaluate the approach on method-level real-bug detection tasks from Defects4J and show that mechanism-guided synthetic-bug feedback improves real-bug detection over execution-, coverage-, mutation-, knowledge-, and search-based baselines. The results suggest that organizing real-bug mechanisms as retrievable and executable feedback targets is an effective way to guide generated tests toward bug-triggering inputs and behavioral oracles.
The asymmetry between language production and perception has been well-documented in psycholinguistics. Whether large language models (LLMs) exhibit a functionally analogous distinction remains an open question, particularly given that LLMs rely on the same underlying mechanism (next-token prediction) for both input and output processing. In this exploratory study, we operationalize the production-perception distinction through direct token probability measurements rather than metalinguistic prompting. Using the base Llama-3.1-8B model, we generated poems under a production prompt and re-scored the same tokens under both rephrased production prompts and perception-oriented prompts. Across an extended experiment with four production and three perception prompts, production-perception distances consistently and substantially exceeded production-production distances, with non-overlapping ranges across conditions and an overall average ratio of approximately 1.8. Near-ceiling correlations in the production-production control confirm that the effect is specific to communicative framing rather than prompt surface variation, and we show the effect replicates across five open-weight models (Llama-3.1-8B, EuroLLM-9B, gemma-2-9b-it, Mistral-7B-Instruct-v0.3, and Qwen2.5-7B-Instruct), spanning both base and instruction-tuned variants. Temporal analysis revealed that the perception prompt exerts its strongest influence at the beginning of the sequence, with divergence decaying as generated context accumulates, though the specific shape of this decay varies across prompt pairs. These findings suggest that prompt framing alone induces a production-perception distinction in LLM probability distributions, even within a decoder-only architecture.
Large Language Models (LLMs) can make exploratory search more efficient but may undermine the reflection and iterative sensemaking needed in unfamiliar domains. Existing LLM tools often prioritize rapid answers over supporting users in tracking how their understanding evolves and how well their strategies align with their goals. We present TrailLM, a system that helps users reconstruct and revisit their exploration paths to support reflection and metacognitive engagement during information seeking. By aligning LLM assistance with users' sensemaking workflows, TrailLM aims to preserve the benefits of LLM-based search while enhancing opportunities for critical reflection on one's own search process.
Existing studies of LLM-as-judge scoring bias work predominantly at the input-output level: they perturb inputs, measure score deltas, and propose prompt-level mitigations. We argue that the same biases admit a representation-level account in the judge's hidden state, complementary to the input-output view and operationally useful in ways it does not afford. We report three findings, across seven judges, seven bias types, and nine benchmarks. Geometry: baseline judging inputs occupy a tight activation manifold while biased inputs are displaced along a low-dimensional, type-specific subspace that sharpens with depth and is recovered consistently by three families of estimators. Causal control: steering hidden states along this subspace drives scoring in both directions, forward shifts reproducing biased scoring on clean inputs and reverse shifts restoring baseline scoring on biased ones, while matched-norm random directions produce shifts an order of magnitude smaller. Operational: a simple linear projection onto the same bias-direction features anticipates judge failures on three entirely unseen benchmarks, substantially outperforming text-based alternatives. Reading bias as activation geometry, rather than as input-output noise, unifies geometric structure, causal control, and operational prediction within a single framework. The project page is available at https://xzx34.github.io/unfair-judge/
Large language models (LLMs) in financial applications fail most consequentially when they are confidently wrong. Hedged, uncertain answers invite scrutiny, whereas confident errors silently degrade downstream decisions without warning. We ask how reliably such confidently wrong answers, or confident hallucinations, can be detected from a model's internal activations, and whether those activations carry information beyond its observable outputs. We train linear probes on the residual stream and evaluate them on two established question-answering (QA) benchmarks built from real filings, FinQA and TAT-QA. Behavioral confidence is measured as the agreement among eight resampled answers to the same question, and probe effectiveness is compared against baselines, such as token log-probabilities and the model's own True/False self-assessment of its answer. Our findings show that among confident answers, those for which all eight resamples agree, 15-23% are wrong on FinQA. There the probes have a significant advantage over baseline methods in detecting hallucinations, holding 0.68-0.77 AUROC while the best baselines fall to 0.55-0.63, across Qwen3-8B, Llama-3.1-8B, and Gemma-2-9B. Our results suggest that probing can be a cost-effective triage mechanism for routing LLM answers to human review and quality control procedures in high-stakes financial applications.
When LLMs exhibit uneven performance across planning tasks, these gaps are often attributed to task difficulty. We argue that this explanation is incomplete, as task-level variation may reflect distinct latent planning competencies rather than differences along a single ability spectrum. We study this question on ACPBench-Hard by evaluating multiple LLM families under varying test-time reasoning budgets and applying a multidimensional item response theory model to uncover the latent competency structure underlying LLM planning. The analysis reveals two principal dimensions that shape planning performance: operational reasoning, the ability to evaluate local action applicability and immediate state transitions, and structural enumeration, the ability to reason about goal reachability and landmark structure. Operational reasoning improving under model scaling and longer reasoning traces, while structural enumeration remains comparatively insensitive. Our findings motivate competency-level evaluation of LLM planning, shifting the focus from whether models improve overall to which planning competencies improve, under what conditions, and why.
A theoretical understanding of Transformers is crucial to better understand the capacities and limitations of large language models (LLMs). There is much work analyzing the expressivity of attention-based models. By proposing handcrafted weights or using computational complexity arguments, a large amount of past theoretical works have sought to characterize which tasks are and which are not in the hypothesis class of Transformer models. However, little work investigates the learnability of such solutions. In this work, we make progress towards this goal. Inspired by recent loss landscape analysis work, we propose preliminary sample complexity bounds for learning C-RASP constructions with Transformers.
Recent text-to-CAD approaches have shown promising results by leveraging large language models, but they often struggle with maintaining structural consistency in complex designs and accurately grounding geometric parameters. To address these issues, we propose HierCAD, a hierarchical text-to-CAD framework that improves both structural reasoning and parameter prediction. HierCAD reformulates CAD generation as progressive reasoning by decomposing CAD construction trees into object-level procedural reasoning and part-level topology reasoning trajectories. To further improve generation fidelity, we introduce a unified Structure Alignment and Parameter Grounding (SAPG) learning strategy. Structure alignment aligns topology reasoning trajectories with their corresponding parametric CAD spans, while parameter grounding mitigates shortcut learning through structure-preserving parameter perturbations and ranking-based supervision. Experiments demonstrate that HierCAD outperforms prior state-of-the-art methods on both CAD sequence generation and reconstructed CAD model evaluation. Our code is available at https://github.com/Collab-Gen/HierCAD.
Cryptocurrency markets are vulnerable to trade-based manipulation, such as wash trading, which can distort price signals and mislead investors. Prior research has mainly focused on detecting manipulation using fixed rules or labeled examples, offering limited flexibility and interpretability for assessing potential risks. Existing visual analytics tools can reveal basic manipulation-related signals, such as token distribution, but still require substantial manual effort to integrate holder relationships, suspicious behaviors, and market dynamics for risk assessment. To address these limitations, we propose ManiScope, an LLM-assisted visual analytics system for analyzing trade-based manipulation risks in cryptocurrency markets. ManiScope provides coordinated views of token distributions, holder relationships, detailed holder behaviors, price dynamics, and suspicious trading patterns. To further enhance user analysis, ManiScope introduces a human-LLM collaborative visual analytics framework. Rather than acting as a basic reactive LLM assistant, the framework positions the LLM as a co-analyst that infers users' analytical intent and emerging hypotheses from interaction context and surfaces relevant visual, statistical, and synthesized evidence for hypothesis evaluation. This design reduces repetitive inspection and strengthens evidence-based reasoning. We evaluate ManiScope through two case studies and a user study with 12 experienced cryptocurrency practitioners. The results suggest that ManiScope supports effective risk assessment of manipulation, reduces manual effort in evidence-seeking, and organizes findings around user hypotheses.
Post-training is essential for refining the domain-specific capabilities of large language models (LLMs), yet existing reward optimization and distribution matching methods tightly couple policy exploration with distribution alignment. This coupling forces expensive exploration directly on the policy model and severely hinders the asynchronous generation, reuse, and cross-model transfer of optimization signals. In this paper, we propose Proxy-guided Update Signal Transfer (PUST), a novel post-training framework that fundamentally decouples update-signal exploration from distribution alignment. Instead of utilizing the primary model for costly exploration, PUST employs a lightweight proxy model as an efficient testbed to discover high-reward behaviors. We extract the relative improvement signal between the proxy's initial and optimized states, transferring this directional update to the primary model to guide its policy alignment. This decoupled pipeline, comprising proxy exploration, update-signal extraction, and signal transfer, significantly reduces computational overhead and enables optimization signals to be asynchronously generated, cached, and reused. Crucially, by transferring relative improvements rather than absolute policy distributions, PUST naturally supports weak-to-strong improvement and seamless cross-model transfer. Systematic evaluations on Qwen3-family models across math and code domains demonstrate that update signals extracted from substantially weaker proxies can robustly and adjustably enhance stronger primary models. Ultimately, PUST transforms post-training from a monolithic online optimization process into a highly modular, reusable, and cost-efficient paradigm.
As Large Language Models (LLMs) are increasingly deployed as conversational tutors, they risk institutionalizing systemic inequalities. This study presents a systematic API audit of four LLMs acting as history tutors, evaluating 1,800 responses regarding the 1989 Romanian Revolution across five student personas varying by ethnicity and socio-economic tier. We uncover four interconnected patterns of \emph{epistemic paternalism}: (1)~\textbf{Differential Refusal}, where safety-aligned models block 76.7\% of educational requests from low-tier students; (2)~\textbf{Epistemic Gatekeeping}, evidenced by a 3$\times$ reduction in access to geopolitical complexity (e.g., the contested ``coup theory'') for marginalized learners; (3)~\textbf{Agency Theft}, a lexical shift where models like LLaMA produce a 5$\times$ higher victimization-to-politics vocabulary ratio for Roma students compared to elite peers; and (4)~\textbf{Elite Hermeneutics}, where AI tutors disproportionately withhold epistemic confidence and justification scores from low-resource demographic profiles. We argue that current safety alignment acts as a paternalistic filter, transforming conversational AI into agents of narrative segregation -- a manifestation of \emph{hermeneutical injustice} in Fricker's~\cite{fricker2007} sense that demands urgent pedagogical auditing.
Large Language Models (LLMs) have achieved remarkable success in complex reasoning tasks through Chain-of-Thought (CoT) prompting. However, these models often exhibit "computational overthinking," generating redundant reasoning steps that increase latency and cost without improving accuracy. Recent studies suggest that CoT trajectories can be significantly pruned, yet existing methods often rely on forcing a static thinking budget, heuristic filtering, sub-optimal early exit via classification, or expensive re-training. In this paper, we introduce OS-Pruner, a lightweight plug-in framework that formulates chain-of-thought pruning as an optimal stopping problem. Given a reasoning prefix, OS-Pruner learns whether further reasoning is worth its token cost by optimizing an explicit utility that trades off final-answer accuracy against generated length. Our novel formulation enables the model to dynamically assess the sufficient point of termination for a reasoning chain. OS-Pruner is designed to be lightweight during both training and inference, and to provide users with fine-grained control over the reasoning-effort vs. accuracy trade-off. On diverse reasoning benchmarks and base models, OS-Pruner achieves 20-60\% reduction in generation length with minimal accuracy sacrifice.
Extending the context length of large language models (LLMs) is critical for many real-world applications, yet standard transformers remain constrained by quadratic compute and linear memory scaling. In this work, we investigate the Associative Recurrent Memory Transformer (ARMT) as a practical approach for enabling long-context processing in LLMs, constant memory scaling, and better efficiency. We make three main contributions. First, we construct two domain-specific long-context datasets designed to evaluate realistic workloads, focusing on narrow-domain fine-tuning scenarios. Second, we propose a comprehensive training recipe for ARMT-based context extension, combining continued pre-training, synthetic long-context data generation, curriculum learning, and selective integration of associative memory into chosen model layers. Third, we present an extensive experimental study demonstrating that ARMT-augmented models: (i) process inputs well beyond their original context limits without degrading performance relative to in-limit baselines; (ii) generalize more effectively to out-of-distribution context lengths; and (iii) need 30% less FLOPs while preserving baseline performance within the original context window.
Recent multimodal large language models (MLLMs) have made remarkable progress on fine-grained perception tasks under the "Thinking with Images" (TwI) paradigm by iteratively performing various visual tool operations. However, this paradigm relies heavily on frequent external tool calls and repeated image re-encoding, which leads to substantial computational overhead and inference latency. To address these issues, we propose Beyond the Eye (BEE), a novel implicit visual tool paradigm centered on self-regulated capability. BEE directly incorporates visual tool invocation behaviors into the training objective and encourages the model to develop a self-regulated invocation mechanism. This design enables the model to adaptively balance internal knowledge and implicit tools, avoiding redundant tool usage while substantially reducing inference latency. Specifically, BEE involves a two-stage training process: (1) Formalized Chain-of-Thought (CoT) Supervised Fine-tuning (SFT). We construct CoT trajectories with structured tool slots and mixed invocation states. This stage activates the model's implicit tool representations and adaptive switching capability. (2) Self-regulated Reward-Driven Alignment. To address redundant tool usage caused by ambiguous cognitive boundaries, we first introduce the Net Tool Gain (NTG) metric to quantify this phenomenon. Based on this observation, we further propose a self-regulated reward mechanism. This mechanism penalizes ineffective tool dependency and encourages the model to perform knowledge routing, ensuring that implicit tools are invoked only when the model's internal knowledge is insufficient. BEE achieves state-of-the-art performance in fine-grained visual perception while remaining competitive in general reasoning tasks and achieving substantial gains in inference efficiency.