Computation and Language
Learning Agentic Policy from Action Guidance
Yuxiang Ji, Zengbin Wang, Yong Wang et al. (+6 more)
Agentic reinforcement learning (RL) for Large
Language Models (LLMs) critically depends on the exploration capability of the base policy, as training signals emerge only within its in-capability region. For tasks where the base policy cannot reach reward states, additional training or external guidance is needed to recover effective learning signals. Rather than relying on costly iterative supervised fine tuning (SFT), we exploit the abundant action data generated in everyday human interactions. We propose ActGuide-RL, which injects action data as plan-style reference guidance, enabling the agentic policy to overcome reachability barriers to reward states. Guided and unguided rollouts are then jointly optimized via mixed-policy training, internalizing the exploration gains back into the unguided policy. Motivated by a theoretical and empirical analysis of the benefit-risk trade-off, we adopt a minimal intervention principle that invokes guidance only as an adaptive fallback, matching task difficulty while minimizing off-policy risk. On search-
agent benchmarks, ActGuide-RL substantially improves over zero RL (+10.7 pp on
GAIA and +19 pp on XBench with Qwen3-4B), and performs on par with the SFT+RL pipeline without any cold start. This suggests a new paradigm for agentic RL that reduces the reliance on heavy SFT data by using scalable action guidance instead.
Read the full abstract ▾
All authors Yuxiang Ji, Zengbin Wang, Yong Wang, Shidong Yang, Ziyu Ma, Guanhua Chen, Zonghua Sun, Liaoni Wu, Xiangxiang Chu
Computation and Language
Towards Visually-Guided Movie Subtitle Translation for Indic Languages
Tarun Chintada, Kshetrimayum Boynao Singh and Asif Ekbal
Movie subtitle translation is inherently
multimodal, yet text-only systems often miss visual cues needed to convey
emotion, action, and social nuance, especially for low-resource Indic
languages (English to Hindi, Bengali, Telugu, Tamil and Kannada). We present a case study on five full-length films and compare two lightweight visual grounding strategies: structured attribute summaries from a 5-minute sliding window and free-text summaries of inter-subtitle visual gaps. Our analysis shows that temporal misalignment between subtitles and frames is a major obstacle in long-form
video, often rendering indiscriminate visual grounding ineffective. However, oracle selective grounding, which replaces only the lowest-quality 20-30% of baseline segments with visual-enhanced outputs, consistently improves COMET over the text-only baseline while requiring far less visual processing. Among the two approaches, coarse attribute-based visual context summarization is more robust, capturing scene-level
emotion and contextual subtle cues that text alone often misses
Read the full abstract ▾
Computation and Language
On Predicting the Post-training Potential of Pre-trained LLMs
Xiaoyuan Li, Yubo Ma, Kexin Yang et al. (+5 more)
The performance of Large
Language Models (LLMs) on downstream tasks is fundamentally constrained by the capabilities acquired during pre-training. However, traditional
benchmarks like MMLU often fail to reflect a base model's plasticity in complex open-ended scenarios, leading to inefficient model selection. We address this by introducing a new task of predicting post-training potential -
forecasting a base model's performance before post-training. We propose RuDE (Rubric-based Discriminative Evaluation), a unified framework that bypasses the generation gap of base models by leveraging response discrimination. Guided by our systematic 4C Taxonomy, RuDE constructs controlled contrastive pairs across diverse domains by fine-grained rubric violations. Extensive experiments demonstrate a correlation greater than 90% with post-training performance. Crucially, validation via Reinforcement Learning (RL) confirms that RuDE effectively identifies high-potential smaller models that outperform larger counterparts, offering a compute-efficient mechanism for foundation model development.
Read the full abstract ▾
All authors Xiaoyuan Li, Yubo Ma, Kexin Yang, Moxin Li, Keqin Bao, Wenie Wang, Fuli Feng, Dayiheng Liu
Computation and Language
Enhancing Target-Guided Proactive Dialogue Systems via Conversational Scenario Modeling and Intent-Keyword Bridging
Maodong Li, Yancui Li and Fang Kong
A target-guided proactive dialogue system aims to steer conversations proactively toward pre-defined targets, such as designated keywords or specific topics. During guided conversations, dynamically modeling conversational scenarios and intent keywords to guide system utterance generation is beneficial; however, existing work largely overlooks this aspect, resulting in a mismatch with the dynamics of real-world conversations. In this paper, we jointly model user profiles and domain knowledge as conversational scenarios to introduce a scenario bias that dynamically influences system utterances, and employ intent-keyword bridging to predict intent keywords for upcoming dialogue turns, providing higher level and more flexible guidance. Extensive automatic and human evaluations demonstrate the effectiveness of conversational scenario modeling and intent keyword bridging, yielding substantial improvements in proactivity, fluency, and informativeness for target-guided proactive dialogue systems, thereby narrowing the gap with real world interactions.
Read the full abstract ▾
Computation and Language · Cross-listed from cs.CV
Maham Nazir, Muhammad Aqeel, Richong Zhang et al. (+1 more)
Multimodal video summarization requires visual features that align semantically with
language generation. Traditional approaches rely on CNN features trained for object classification, which represent visual concepts as discrete categories not aligned with natural
language. We propose ClipSum, a framework that leverages frozen CLIP
vision-
language features with explicit temporal modeling and dimension-adaptive fusion for instructional
video summarization. CLIP's contrastive pre-training on 400M image-text pairs yields visual features semantically aligned with the linguistic concepts that text decoders generate, bridging the
vision-
language gap at the representation level. On YouCook2, ClipSum achieves 33.0% ROUGE-1 versus 30.5% for ResNet-152 with 4x lower dimensionality (512 vs. 2048), demonstrating that semantic
alignment matters more than feature capacity. Frozen CLIP (33.0%) surpasses fine-tuned CLIP (32.3%), showing that preserving pre-trained
alignment is more valuable than task-specific adaptation. https://github.com/aqeeelmirza/clipsum
Read the full abstract ▾
All authors Maham Nazir, Muhammad Aqeel, Richong Zhang, Francesco Setti
Computation and Language · Cross-listed from cs.SE
StepCodeReasoner: Aligning Code Reasoning with Stepwise Execution Traces via Reinforcement Learning
Hao Wang, Rui Li, Lei Sha et al. (+1 more)
Existing code
reasoning methods primarily supervise final code outputs, ignoring intermediate states, often leading to reward hacking where correct answers are obtained through inconsistent
reasoning. We propose StepCodeReasoner, a framework that introduces explicit intermediate execution-state supervision. By automatically inserting structured print-based execution-trace anchors into code, the model is trained to predict runtime states at each step, transforming code
reasoning into a verifiable, stepwise execution modeling problem. Building on this execution-aware method, we introduce Bi-Level GRPO, a reinforcement learning
algorithm for structured credit assignment at two levels: inter-trajectory, comparing alternative execution paths, and intra-trajectory, rewarding intermediate accuracy based on its impact on downstream correctness. Extensive experiments demonstrate that StepCodeReasoner achieves SOTA performance in code
reasoning. In particular, our 7B model achieves 91.1% on CRUXEval and 86.5% on LiveCodeBench, outperforming the CodeReasoner-7B baseline (86.0% and 77.7%) and GPT-4o (85.6% and 75.1%). Furthermore, on the execution-trace
benchmark REval, our model scores 82.9%, outperforming baseline CodeReasoner-7B (72.3%), its 14B counterpart (81.1%), and GPT-4o (77.3%). Additionally, our approach also improves code generation performance, demonstrating that explicit execution modeling enhances both code
reasoning and code generation.
Read the full abstract ▾
All authors Hao Wang, Rui Li, Lei Sha, Jie M. Zhang
Computation and Language
YFPO: Yoked Feature Preference Optimization with Neuron-Guided Rewards
Yifan Le
Preference optimization has become a widely used post-training paradigm for improving the
reasoning abilities of large
language models. Existing methods typically learn from preferred and dispreferred responses as external behavioral supervision, while largely ignoring capability-related signals encoded in the model's internal representations. In this work, we study whether such internal signals can provide useful auxiliary supervision for mathematical
reasoning. We introduce YFPO (Yoked Feature Preference Optimization), a
neuron-guided preference optimization framework that couples response-level preference learning with
neuron-level rewards. YFPO first uses AttnLRP to identify math-associated internal features, and then derives an auxiliary reward from the activation margin of these
neurons between preferred and dispreferred responses. This reward is combined with the standard preference optimization objective, encouraging the model to align external preferences with internal math-related features. We conduct small-scale experiments on GSM8K with a compact
language model. Results show that
neuron-guided rewards can influence preference optimization dynamics and yield measurable improvements in several settings, suggesting that internal representations can serve as lightweight and interpretable signals for
reasoning-oriented post-training.
Read the full abstract ▾
Computation and Language
Qwen-Scope: Turning Sparse Features into Development Tools for Large Language Models
Boyi Deng, Xu Wang, Yaoning Wang et al. (+15 more)
Large
language models have achieved remarkable capabilities across diverse tasks, yet their internal decision-making processes remain largely opaque, limiting our ability to inspect, control, and systematically improve them. This opacity motivates a growing body of research in mechanistic interpretability, with sparse autoencoders (SAEs) emerging as one of the most promising tools for decomposing model activations into sparse, interpretable feature representations. We introduce Qwen-Scope, an open-source suite of SAEs built on the Qwen model family, comprising 14 groups of SAEs across 7 model variants from the Qwen3 and Qwen3.5 series, covering both dense and mixture-of-expert architectures. Built on top of these SAEs, we show that SAEs can go beyond post-hoc analysis to serve as practical interfaces for model development along four directions: (i) inference-time steering, where SAE feature directions control
language, concepts, and preferences without modifying model weights; (ii) evaluation analysis, where activated SAE features provide a representation-level proxy for
benchmark redundancy and capability coverage; (iii) data-centric workflows, where SAE features support
multilingual toxicity classification and safety-oriented data synthesis; and (iv) post-training optimization, where SAE-derived signals are incorporated into supervised fine-tuning and reinforcement learning objectives to mitigate undesirable behaviors such as code-switching and repetition. Together, these results demonstrate that SAEs can serve not only as post-hoc analysis tools, but also as reusable representation-level interfaces for diagnosing, controlling, evaluating, and improving large
language models. By open-sourcing Qwen-Scope, we aim to support mechanistic research and accelerate practical workflows that connect model internals to downstream behavior.
Read the full abstract ▾
All authors Boyi Deng, Xu Wang, Yaoning Wang, Yu Wan, Yubo Ma, Baosong Yang, Haoran Wei, Jialong Tang, Huan Lin, Ruize Gao, Tianhao Li, Qian Cao, Xuancheng Ren, Xiaodong Deng, An Yang, Fei Huang, Dayiheng Liu, Jingren Zhou
Computation and Language
Concordance Comparison as a Means of Assembling Local Grammars
Juliana Pirovani, Elias de Oliveira and Eric Laporte
Named Entity Recognition for person names is an important but non-trivial task in information extraction. This article uses a tool that compares the concordances obtained from two local grammars (LG) and highlights the differences. We used the results as an aid to select the best of a set of LGs. By analyzing the comparisons, we observed relationships of inclusion, intersection and disjunction within each pair of LGs, which helped us to assemble those that yielded the best results. This approach was used in a case study on extraction of person names from texts written in Portuguese. We applied the enhanced grammar to the Gold Collection of the Second HAREM. The F-Measure obtained was 76.86, representing a gain of 6 points in relation to the state-of-the-art for Portuguese.
Read the full abstract ▾
Computation and Language · Cross-listed from cs.CV
UniVLR: Unifying Text and Vision in Visual Latent Reasoning for Multimodal LLMs
Houcheng Jiang, Jiajun Fu, Junfeng Fang et al. (+4 more)
Multimodal large
language models are increasingly expected to perform thinking with images, yet existing visual latent
reasoning methods still rely on explicit textual chain-of-thought interleaved with visual latent
tokens. This interleaved design limits efficiency and keeps
reasoning fragmented across separate text and
vision channels. We propose UniVLR, a unified visual latent
reasoning framework that treats textual
reasoning and auxiliary visual evidence as a shared visual workspace. Instead of preserving text CoT as an independent inference-time path, UniVLR renders
reasoning traces together with auxiliary images and learns to compress this unified representation into compact visual latent
tokens. At inference time, the model reasons only through visual latents and directly decodes the final answer, avoiding both external tool calls and verbose text
reasoning. Experiments on real-world perception and visual
reasoning tasks show that UniVLR outperforms prior visual latent
reasoning methods while using substantially fewer generated
reasoning tokens, suggesting a more unified and efficient paradigm for visual thinking in MLLMs.
Read the full abstract ▾
All authors Houcheng Jiang, Jiajun Fu, Junfeng Fang, Chen Gao, Xiang Wang, Xiangnan He, Yong Li
Computation and Language
Self-Distilled Trajectory-Aware Boltzmann Modeling: Bridging the Training-Inference Discrepancy in Diffusion Language Models
Kecheng Chen, Ziru Liu, Xijia Tao et al. (+9 more)
Diffusion Language Models (DLMs) have recently emerged as a promising alternative to autoregressive
language models, offering stronger global awareness and highly parallel generation. However, post-training DLMs with standard Negative Evidence Lower Bound (NELBO)-based supervised fine-tuning remains inefficient: training reconstructs randomly masked
tokens in a single step, whereas inference follows a confidence-guided, multi-step easy-to-hard denoising trajectory. Recent trajectory-based self-
distillation methods exploit such inference trajectories mainly for sampling-step compression and acceleration, often improving decoding efficiency without substantially enhancing the model's underlying capability, and may even degrade performance under full
diffusion decoding. In this work, we ask whether self-distilled trajectories can be used not merely for faster inference, but for genuine knowledge acquisition. Although these trajectories lie on the pretrained DLM's own distributional
manifold and thus offer a potentially lower optimization barrier, we find that naively fine-tuning on them with standard NELBO objectives yields only marginal gains. To address this limitation, we propose Trajectory-Aligned optimization via Boltzmann Modeling (TABOM), a self-distilled trajectory-based post-training framework that aligns training with the easy-to-hard structure of inference. TABOM models the inference unmasking preference as a Boltzmann distribution over predictive
entropies and derives a tractable pairwise ranking objective to align the model's certainty ordering with the observed decoding trajectory. Empirically, TABOM achieves substantial gains in new domains, expands the effective knowledge boundary of DLMs, and significantly mitigates catastrophic forgetting compared with standard SFT.
Read the full abstract ▾
All authors Kecheng Chen, Ziru Liu, Xijia Tao, Hui Liu, Yibing Liu, Xinyu Fu, Shi Wu, Suiyun Zhang, Dandan Tu, Lingpeng Kong, Rui Liu, Haoliang Li
Computation and Language · Cross-listed from cs.LG
GEAR: Granularity-Adaptive Advantage Reweighting for LLM Agents via Self-Distillation
Sijia Li, Yuchen Huang, Zifan Liu et al. (+7 more)
Reinforcement learning has become a widely used post-training approach for LLM
agents, where training commonly relies on outcome-level rewards that provide only coarse supervision. While finer-grained credit assignment is promising for effective policy updates, obtaining reliable local credit and assigning it to the right parts of the long-horizon trajectory remains an open challenge. In this paper, we propose Granularity-adaptivE Advantage Reweighting (GEAR), an adaptive-granularity credit assignment framework that reshapes the trajectory-level GRPO advantage using
token- and segment-level signals derived from self-
distillation. GEAR compares an on-policy student with a ground-truth-conditioned teacher to obtain a reference-guided divergence signal for identifying adaptive segment boundaries and modulating local advantage weights. This divergence often spikes at the onset of a semantic deviation, while later
tokens in the same autoregressive continuation may return to low divergence. GEAR therefore treats such spikes as anchors for adaptive credit regions: where the student remains aligned with the teacher,
token-level resolution is preserved; where it departs, GEAR groups the corresponding continuation into an adaptive segment and uses the divergence at the departure point to modulate the segment' s advantage. Experiments across eight mathematical
reasoning and agentic tool-use
benchmarks with Qwen3 4B and 8B models show that GEAR consistently outperforms standard GRPO, self-
distillation-only baselines, and
token- or turn-level credit-assignment methods. The gains are especially strong on
benchmarks with lower GRPO baseline accuracy, reaching up to around 20% over GRPO, suggesting that the proposed adaptive reweighting scheme is especially useful in more challenging long-horizon settings.
Read the full abstract ▾
All authors Sijia Li, Yuchen Huang, Zifan Liu, Yanping Li, Jingjing Fu, Li Zhao, Jiang Bian, Ling Zhang, Jun Zhang, Rui Wang
Computation and Language
Probabilistic Calibration Is a Trainable Capability in Language Models
Davide Baldelli, Sruthi Kuriakose, Maryam Hashemzadeh et al. (+2 more)
Language models are increasingly used in settings where outputs should follow an intended distribution, yet their generation probabilities are often poorly calibrated to it. In many such requests the distribution is implicit and the valid outputs cannot be listed in advance, so the task cannot be passed to an external random number generator. We study whether this capability can be improved directly through fine-tuning. Concretely, we fine-tune
language models on synthetic
prompts that require sampling from mathematical distributions, and compare two Calibration Fine-Tuning variants: a soft-target method that converts the desired output distribution into trie-derived next-
token targets, and a hard-target method that trains on sampled completions from the same target distribution. Across 12 models spanning four families, both methods substantially improve structured-sampling fidelity on held-out distribution families and unseen parameter settings, showing that training on cheap synthetic mathematical distributions improves sampling on unseen distributions, even in natural
language. The gains also transfer to broader
stochastic generation
benchmarks, including open-ended random generation, where fine-tuned models broaden their first-
token support by one to two orders of magnitude, and NoveltyBench. The gains sometimes reduce downstream capability, especially arithmetic
reasoning, with costs varying by model. Overall, our results show that probabilistic calibration can be improved through fine-tuning with either objective, and that calibration learned on mathematical distributions transfers to natural-
language stochastic generation. Code is available at https://github.com/chandar-lab/calibration-finetuning.
Read the full abstract ▾
All authors Davide Baldelli, Sruthi Kuriakose, Maryam Hashemzadeh, Amal Zouaq, Sarath Chandar
Computation and Language · Cross-listed from cs.LG
More Edits, More Stable: Understanding the Lifelong Normalization in Sequential Model Editing
Xin Ma, Wei Chen, Qi Liu et al. (+4 more)
Lifelong Model Editing aims to continuously update evolving facts in Large
Language Models while preserving unrelated knowledge and general capabilities, yet it remains plagued by catastrophic forgetting and model collapse. Empirically, we find that recent editors resilient over long horizons share the same core strategy: Lifelong Normalization (LN), which normalizes value gradients using running statistics. Removing LN causes immediate performance collapse, and we observe a counter-intuitive positive cumulative effect where early edits can promote the success of future edits. Yet the mechanism of LN remains a "black box", leaving its precise role in lifelong stability poorly understood. In this work, we provide the first theoretical account of LN in the lifelong regime. Our analysis reveals a self-reinforcing stability loop and proves that, when combined with ridge-regularized regression, LN yields parameter updates with asymptotic orthogonality and bounded norms, directly mitigating forgetting and systemic collapse. Based on these insights, we derive StableEdit, which strengthens this stability loop via an explicit warm-up stage and full whitening, improving long-horizon stability at minimal overhead. Extensive experiments validate our theory and demonstrate competitive performance. Our code is available at https://github.com/MINE-USTC/StableEdit.
Read the full abstract ▾
All authors Xin Ma, Wei Chen, Qi Liu, Derong Xu, Zhi Zheng, Tong Xu, Enhong Chen
Computation and Language · Cross-listed from cs.LG
ROMER: Expert Replacement and Router Calibration for Robust MoE LLMs on Analog Compute-in-Memory Systems
Wenyong Zhou, Yuannuo Feng, Yizhe Chen et al. (+6 more)
Large
language models (LLMs) with mixture-of-experts (MoE) architectures achieve remarkable scalability by sparsely activating a subset of experts per
token, yet their frequent expert switching creates
memory bandwidth bottlenecks that compute-in-
memory (CIM) architectures are well-suited to mitigate. However, analog CIM systems suffer from inherent hardware imperfections that perturb stored weights, and its negative impact on MoE-based LLMs in noisy CIM environments remains unexplored. In this work, we present the first systematic investigation of MoE-based LLMs under noise model calibrated with real chip measurements, revealing that hardware noise critically disrupts expert load balance and renders clean-trained routing decisions consistently suboptimal. Based on these findings, we propose ROMER, a post-training calibration framework that (1) replaces underactivated experts with high-frequency ones to restore load balance, and (2) recalibrates router logits via percentile-based normalization to stabilize routing under noise. Extensive experiments across multiple
benchmarks demonstrate that ROMER achieves up to 58.6%, 58.8%, and 59.8% reduction in perplexity under real-chip noise conditions for DeepSeek-MoE, Qwen-MoE, and OLMoE, respectively, establishing its effectiveness and generalizability across diverse MoE architectures.
Read the full abstract ▾
All authors Wenyong Zhou, Yuannuo Feng, Yizhe Chen, Taiqiang Wu, Wendong Xu, Wenbo Qi, Zhengwu Liu, Wang Kang, Ngai Wong
Computation and Language
Choosing features for classifying multiword expressions
Eric Laporte
Multiword expressions (MWEs) are a heterogeneous set with a glaring need for classifications. Designing a satisfactory classification involves choosing features. In the case of MWEs, many features are a priori available. Not all features are equal in terms of how reliably MWEs can be assigned to classes. Accordingly, resulting classifications may be more or less fruitful for computational use. I outline an enhanced classification. In order to increase its suitability for many
languages, I use previous works taking into account various
languages.
Read the full abstract ▾
Computation and Language · Cross-listed from cs.LG
Entropy Polarity in Reinforcement Fine-Tuning: Direction, Asymmetry, and Control
Jiazheng Zhang, Ziche Fu, Junrui Shen et al. (+17 more)
Policy
entropy has emerged as a fundamental measure for understanding and controlling exploration in reinforcement learning with verifiable rewards (RLVR) for LLMs. However, existing
entropy-aware methods mainly regulate
entropy through global objectives, while the
token-level mechanism by which sampled policy updates reshape policy
entropy remains underexplored. In this work, we develop a theoretical framework of
entropy mechanics in RLVR. Our analysis yields a first-order approximation of the
entropy change, giving rise to
entropy polarity, a signed
token-level quantity that predicts how much a sampled update expands or contracts
entropy. This analysis further reveals a structural asymmetry: reinforcing frequent high-probability
tokens triggers contraction tendencies, whereas expansive tendencies typically require lower-probability samples or stronger distributional correction. Empirically, we show that
entropy polarity reliably predicts
entropy changes, and that positive and negative polarity branches play complementary roles in preserving exploration while strengthening exploitation. Building on these insights, we propose Polarity-Aware Policy Optimization (PAPO), which preserves both polarity branches and implements
entropy control through advantage reweighting. With the empirical
entropy trajectory as an online phase signal, PAPO adaptively reallocates optimization pressure between
entropy-expanding and
entropy-contracting updates. Experiments on mathematical
reasoning and agentic
benchmarks show that PAPO consistently outperforms competitive baselines, while delivering superior training efficiency and substantial reward improvements.
Read the full abstract ▾
All authors Jiazheng Zhang, Ziche Fu, Junrui Shen, Yunbin Zhao, Yunke Zhang, Zhiheng Xi, Long Ma, Chenxin An, Zhihao Zhang, Shichun Liu, Dingwei Zhu, Shihan Dou, Shaofan Liu, Han Li, Wiggin Zhou, Aiden Adams, Tao Gui, Fei Huang, Qi Zhang, Xuanjing Huang
Computation and Language
From Token to Token Pair: Efficient Prompt Compression for Large Language Models in Clinical Prediction
Mingcheng Zhu, Zhiyao Luo, Yu Liu et al. (+1 more)
By processing electronic health records (EHRs) as natural
language sequences, large
language models (LLMs) have shown potential in clinical prediction tasks such as mortality prediction and phenotyping. However, longitudinal or highly frequent EHRs often yield excessively long
token sequences that result in high computational costs and even reduced performance. Existing solutions either add modules for compression or remove less important
tokens, which introduce additional inference latency or risk losing clinical information. To achieve lossless compression of
token sequences without additional cost or loss of performance, we propose Medical
Token-Pair Encoding (MedTPE), a layered method that extends standard tokenisation for EHR sequences. MedTPE merges frequently co-occurring medical
token pairs into composite
tokens, providing lossless compression while preserving the computational complexity through a dependency-aware replacement strategy. Only the
embeddings of the newly introduced
tokens of merely 0.5-1.0% of the LLM's parameters are fine-tuned via self-supervised learning. Experiments on real-world
datasets for two clinical scenarios demonstrate that MedTPE reduces input
token length by up to 31% and inference latency by 34-63%, while maintaining or even improving both predictive performance and output format compliance across multiple LLMs and four clinical prediction tasks. Furthermore, MedTPE demonstrates robustness across different input context lengths and generalisability to scientific and financial domains and different
languages.
Read the full abstract ▾
All authors Mingcheng Zhu, Zhiyao Luo, Yu Liu, Tingting Zhu
Computation and Language
Safety-Oriented Evaluation of Language Understanding Systems for Air Traffic Control
Yujing Chang, Yash Guleria, Duc-Thinh Pham et al. (+4 more)
Air Traffic Control (ATC) is a safety-critical domain in which incorrect interpretation of instructions may lead to severe operational consequences. While large
language models (LLMs) demonstrate strong general performance, their reliability in operational ATC environments remains unclear. Existing evaluation approaches, largely based on aggregate metrics such as F1 or macro accuracy, treat all errors uniformly and fail to account for the asymmetric consequences of high-risk semantic mistakes (e.g., incorrect runway identifiers or movement constraints). To address this gap, we propose a safety-oriented, consequence-aware evaluation framework tailored to ATC operations. Our results reveal that while current LLMs achieve reasonable aggregate accuracy, their operational reliability is severely limited. Evaluated on clean transcripts, the peak Risk Score reaches only 0.69, with most models scoring below 0.6 despite high macro-F1 performance. Further analysis shows that errors concentrate in high-impact entities despite relatively stable action-type classification, indicating structural grounding deficiencies. These findings highlight the necessity of consequence-aware evaluation protocols for the responsible deployment of AI-assisted ATC systems.
Read the full abstract ▾
All authors Yujing Chang, Yash Guleria, Duc-Thinh Pham, Nhut-Huy Pham, Ningli Wang, Vu N. Duong, Sameer Alam
Computation and Language · Cross-listed from cs.RO
DreamAvoid: Critical-Phase Test-Time Dreaming to Avoid Failures in VLA Policies
Xianzhe Fan, Yuxiang Lu, Shenyuan Gao et al. (+4 more)
Vision-
Language-Action (VLA) models are often brittle in fine-grained manipulation, where minor action errors during the critical phases can rapidly escalate into irrecoverable failures. Since existing VLA models rely predominantly on successful demonstrations for training, they lack an explicit awareness of failure during these critical phases. To address this, we propose DreamAvoid, a critical-phase test-time dreaming framework that enables VLA models to anticipate and avoid failures. We also introduce an autonomous boundary learning paradigm to refine the system's understanding of the subtle boundary between success and failure. Specifically, we (1) utilize a Dream Trigger to determine whether the execution has entered a critical phase, (2) sample multiple candidate action chunks from the VLA via an Action Proposer, and (3) employ a Dream Evaluator, jointly trained on mixed data (success, failure, and boundary cases), to "dream" the short-horizon futures corresponding to the candidate actions, evaluate their values, and select the optimal action. We conduct extensive evaluations on real-world manipulation tasks and simulation
benchmarks. The results demonstrate that DreamAvoid can effectively avoid failures, thereby improving the overall task success rate. Across four real-world tasks, DreamAvoid achieves 72.5% success, compared with 48.8% for the base policy and 54.4% for GPC-RANK. Our code is available at https://github.com/XianzheFan/DreamAvoid.
Read the full abstract ▾
All authors Xianzhe Fan, Yuxiang Lu, Shenyuan Gao, Xiaoyang Wu, Ruihua Han, Manling Li, Hengshuang Zhao
Computation and Language
Training-Inference Consistent Segmented Execution for Long-Context LLMs
Xianpeng Shang, Jiang Li, Zehua Duo et al. (+2 more)
Transformer-based large
language models face severe scalability challenges in long-context generation due to the computational and
memory costs of full-context
attention. Under practical computation and
memory constraints, many inference-efficient long-context methods improve efficiency by adopting bounded-context or segment-level execution only during inference, while continuing to train models under full-context
attention, resulting in a mismatch between training and inference execution and state-transition semantics. Based on this insight, we propose a training-inference consistent segment-level generation framework, in which training and inference follow the same segment-level forward execution semantics. During training, consistency with inference is enforced by restricting gradient propagation to KV states carried over from the immediately preceding segment, while permitting head-specific access to past KV states during the forward pass without involving them in gradient propagation. Across long-context
benchmarks, our approach achieves performance comparable to full-context
attention, while achieving competitive latency-
memory trade-offs against strong inference-efficient baselines, and substantially improving scalability at very long context lengths (e.g., approximately 6x lower peak prefill
memory at 128K compared to full-context
attention with FlashAttention).
Read the full abstract ▾
All authors Xianpeng Shang, Jiang Li, Zehua Duo, Qianyi Cai, Xiangdong Su
Computation and Language
Learning to Foresee: Unveiling the Unlocking Efficiency of On-Policy Distillation
Yuchen Cai, Ding Cao, Liang Lin et al. (+9 more)
On-policy
distillation (OPD) has emerged as an efficient post-training paradigm for large
language models. However, existing studies largely attribute this advantage to denser and more stable supervision, while the parameter-level mechanisms underlying OPD's efficiency remain poorly understood. In this work, we argue that OPD's efficiency stems from a form of “foresight”: it establishes a stable update trajectory toward the final model early in training. This foresight manifests in two aspects. First, at the Module-Allocation Level, OPD identifies regions with low marginal utility and concentrates updates on modules that are more critical to
reasoning. Second, at the Update-Direction Level, OPD exhibits stronger low-rank concentration, with its dominant subspaces aligning closely with the final update subspace early in training. Building on these findings, we propose EffOPD, a plug-and-play acceleration method that speeds up OPD by adaptively selecting an extrapolation step size and moving along the current update direction. EffOPD requires no additional trainable modules or complex hyperparameter tuning, and achieves an average training acceleration of $3\times$ while maintaining comparable final performance. Overall, our findings provide a parameter-dynamics perspective for understanding the efficiency of OPD and offer practical insights for designing more efficient post-training methods for large
language models.
Read the full abstract ▾
All authors Yuchen Cai, Ding Cao, Liang Lin, Chunxi Luo, Xin Xu, Kai Yang, Weijie Liu, Saiyong Yang, Tianxiang Zhao, Guangzhong Sun, Guiquan Liu, Junfeng Fang
Computation and Language · Cross-listed from cs.IR
AgentDisCo: Towards Disentanglement and Collaboration in Open-ended Deep Research Agents
Jiarui Jin, Zexuan Yan, Shijian Wang et al. (+2 more)
In this paper, we present AgentDisCo, a novel Disentangled and Collaborative agentic architecture that formulates deep research as an
adversarial optimization problem between information exploration and exploitation. Unlike existing approaches that conflate these two processes into a single module, AgentDisCo employs a critic
agent to evaluate generated outlines and refine search queries, and a generator
agent to retrieve updated results and revise outlines accordingly. The iteratively refined outline is then passed to a downstream report writer that synthesizes a comprehensive research report. The overall workflow supports both handcrafted and automatically discovered design strategies via a meta-optimization harness, in which the generator
agent is repurposed as a scoring
agent to evaluate critic outputs and generate quality signals. Powerful code-generation
agents (e.g., Claude-Code, Codex) systematically explore
agent configurations and construct a policy bank, a structured repository of reusable design strategies, enabling the framework to self-refine without extensive human intervention. We evaluate AgentDisCo on three established deep research
benchmarks (DeepResearchBench, DeepConsult, DeepResearchGym) using Gemini-2.5-Pro, achieving performance comparable to or surpassing leading closed-source systems. Observing that existing
benchmarks inadequately reflect real-world user needs, we introduce GALA (General AI Life Assistants), a
benchmark that mines latent research interests from users' historical browsing behavior. We further develop a rendering
agent that converts research reports into visually rich poster presentations, and demonstrate an end-to-end product, AutoResearch Your Interest, which delivers personalized deep research recommendations derived from individual browsing histories.
Read the full abstract ▾
All authors Jiarui Jin, Zexuan Yan, Shijian Wang, Wenxiang Jiao, Yuan Lu
Computation and Language · Cross-listed from cs.AI
Allegory of the Cave: Measurement-Grounded Vision-Language Learning
Kepeng Xu, Li Xu, Gang He et al. (+1 more)
Vision-
language models typically reason over post-ISP RGB images, although RGB rendering can clip, suppress, or quantize sensor evidence before inference. We study whether grounding improves when the visual interface is moved closer to the underlying camera measurement. We formulate measurement-grounded
vision-
language learning and instantiate it as PRISM-VL, which combines RAW-derived Meas.-XYZ inputs, camera-conditioned grounding, and Exposure-Bracketed Supervision Aggregation for transferring supervision from RGB proxies to measurement-domain observations. Using a quality-controlled 150K instruction-tuning set and a held-out
benchmark targeting low-light, HDR, visibility-sensitive, and
hallucination-sensitive cases, PRISM-VL-8B reaches 0.6120 BLEU, 0.4571 ROUGE-L, and 82.66% LLM-Judge accuracy, improving over the RGB Qwen3-VL-8B baseline by +0.1074 BLEU, +0.1071 ROUGE-L, and +4.46 percentage points. These results suggest that part of VLM grounding error arises from information lost during RGB rendering, and that preserving measurement-domain evidence can improve
multimodal reasoning.
Read the full abstract ▾
All authors Kepeng Xu, Li Xu, Gang He, Wenxin Yu
Computation and Language · Cross-listed from cs.LG
Slicing and Dicing: Configuring Optimal Mixtures of Experts
Margaret Li, Sneha Kudugunta, Danielle Rothermel et al. (+1 more)
Mixture-of-Experts (MoE) architectures have become standard in large
language models, yet many of their core design choices - expert count, granularity, shared experts, load balancing,
token dropping - have only been studied one or two at a time over narrow configuration ranges. It remains an open question whether these choices can be optimized independently, without considering interactions. We present the first systematic study of over 2,000 pretraining runs spanning models up to 6.6B total parameters, in which we exhaustively vary total experts, expert dimension, heterogeneous expert sizing within a single layer, shared expert size and load-balancing mechanisms. We find that at every active-parameter scale that we study, performance consistently improves with total MoE parameters even at extreme active expert parameter ratios like 128.Further, the optimal expert size is nearly invariant to total parameter count and depends only on active parameter count. Third, we see that other choices like shared experts, heterogeneous experts and load-balancing settings have small effects relative to expert count and granularity, although dropless routing yields a consistent gain. Overall, our results suggest a simpler recipe: focus on expert count and granularity, other choices have minimal effect on final quality.
Read the full abstract ▾
All authors Margaret Li, Sneha Kudugunta, Danielle Rothermel, Luke Zettlemoyer
Computation and Language
Robust LLM Unlearning Against Relearning Attacks: The Minor Components in Representations Matter
Zeguan Xiao, Xuanzhe Xu, Yong Wang et al. (+3 more)
Large
language model (LLM) unlearning aims to remove specific data influences from pre-trained model without costly retraining, addressing
privacy, copyright, and safety concerns. However, recent studies reveal a critical vulnerability: unlearned models rapidly recover "forgotten" knowledge through relearning attacks. This fragility raises serious security concerns, especially for open-weight models. In this work, we investigate the fundamental mechanism underlying this fragility from a representation geometry perspective. We discover that existing unlearning methods predominantly optimize along dominant components, leaving minor components largely unchanged. Critically, during relearning attacks, the modifications in these dominant components are easily reversed, enabling rapid knowledge recovery, whereas minor components exhibit stronger resistance to such reversal. We further provide a theoretical analysis that explains both observations from the spectral structure of representations. Building on this insight, we propose Minor Component Unlearning (MCU), a novel unlearning approach that explicitly targets minor components in representations. Extensive experiments on three
datasets validate that, by concentrating unlearning effects in these inherently robust directions, our method achieves substantially improved resistance to relearning attacks.
Read the full abstract ▾
All authors Zeguan Xiao, Xuanzhe Xu, Yong Wang, Jian Yang, Yanqing HU, Yun Chen
Computation and Language
Human-Grounded Multimodal Benchmark with 900K-Scale Aggregated Student Response Distributions from Japan's National Assessment of Academic Ability
Kyosuke Takami, Yuka Tateisi, Satoshi Sekine et al. (+1 more)
Authentic school examinations provide a high-validity test bed for evaluating
multimodal large
language models (MLLMs), yet
benchmarks grounded in Japanese K-12 assessments remain scarce. We present a
multimodal dataset constructed from Japan's National Assessment of Academic Ability, comprising officially released middle-school items in Science, Mathematics, and Japanese
Language. Unlike existing
benchmarks based on synthetic or curated data, our
dataset preserves real exam layouts, diagrams, and Japanese educational text, together with nationwide aggregated student response distributions (N $\approx$ 900{,}000). These features enable direct comparison between human and model performance under a unified evaluation framework. We
benchmark recent
multimodal LLMs using exact-match accuracy and character-level F1 for open-ended responses, observing substantial variation across subjects and strong sensitivity to visual
reasoning demands. Human evaluation and LLM-as-judge analyses further assess the reliability of automatic scoring. Our
dataset establishes a reproducible, human-grounded
benchmark for
multimodal educational
reasoning and supports future research on evaluation, feedback generation, and explainable AI in authentic assessment contexts. Our
dataset is available at: https://github.com/KyosukeTakami/gakucho-
benchmark
Read the full abstract ▾
All authors Kyosuke Takami, Yuka Tateisi, Satoshi Sekine, Yusuke Miyao
Computation and Language · Cross-listed from cs.CV
Hide to See: Reasoning-prefix Masking for Visual-anchored Thinking in VLM Distillation
Seonghoon Yu, Dongjun Nam, Byung-Kwan Lee et al. (+1 more)
Recent think-answer approaches in VLMs, such as Qwen3-VL-Thinking, boost
reasoning performance by leveraging intermediate thinking steps before the final answer, but their computational cost becomes substantial, especially for larger VLMs. To distill such capabilities into compact think-answer VLMs, a primary objective is to improve the student's ability to utilize visual evidence throughout its
reasoning trace, as long think-answer traces suffer from visual forgetting issues. To this end, we introduce a novel think-answer
distillation framework that encourages the student to anchor its thinking on visual information by masking the student's salient
reasoning prefixes. To compensate for such masked textual cues, the student is encouraged to rely more on visual evidence as an alternative source of information during
distillation. Our masking strategies include: 1)
token-wise salient
reasoning-prefix masking, which masks high-influence
reasoning prefixes selectively for each next-
token prediction, and 2) self-paced masking budget scheduling, which gradually increases the masking scale according to
distillation difficulty, measured by the discrepancy between teacher–student distributions. In the
distillation phase, the student is guided by our salient
reasoning-prefix mask, which blocks both future
tokens and salient
reasoning cues, in place of the standard causal mask used for auto-regressive
language modeling. Experimental results show that our approach outperforms recent open-source VLMs, VLM
distillation, and self-
distillation methods on
multimodal reasoning benchmarks, while further analyzes confirm enhanced visual utilization along the student thinking process.
Read the full abstract ▾
All authors Seonghoon Yu, Dongjun Nam, Byung-Kwan Lee, Jeany Son
Computation and Language
Macro: Enhancing Multilingual Counterfactual Explanations through Alignment-as-Preference Optimization
Yilong Wang, Qianli Wang, Bohao Chu et al. (+3 more)
Self-generated counterfactual explanations (SCEs) are minimally modified inputs (minimality) generated by large
language models (LLMs) that flip their own predictions (validity), offering a causally grounded approach to unraveling black-box LLM behavior. Yet extending them beyond English remains challenging: existing methods struggle to produce valid SCEs in non-dominant
languages, and a persistent trade-off between validity and minimality undermines explanation quality. We introduce Macro, a preference
alignment framework that applies Direct Preference Optimization (DPO) to
multilingual SCE generation, using a composite scoring function to construct preference pairs that effectively translate the trade-off into measurable preference signals. Experiments across four LLMs and seven typologically diverse
languages show that Macro improves validity by 12.55% on average over the chain-of-thought baseline without degrading minimality, while avoiding the severe minimality violations of the translation-based baseline. Compared to supervised fine-tuning, Macro achieves superior performance on both metrics, confirming that explicit preference optimization is essential for balancing this trade-off. Further analyses reveal that Macro increases cross-lingual perturbation
alignment and mitigates common generation errors. Our results highlight preference optimization as a promising direction for enhancing
multilingual model explanations.
Read the full abstract ▾
All authors Yilong Wang, Qianli Wang, Bohao Chu, Yihong Liu, Jing Yang, Simon Ostermann
Computation and Language
Yuanhao Yue, Chengyu Wang, Yuanjie Lyu et al. (+2 more)
Recent
multimodal large
language models (MLLMs) have shown strong chain-of-thought (CoT)
reasoning ability on
vision-
language tasks, but their direct deployment in real-world systems is often limited by latency and resource constraints. In practice, smaller MLLMs are preferred for online serving, yet their
reasoning performance is bottlenecked by the lack of large-scale, high-quality
multimodal CoT supervision. In this paper, we present OmniThoughtVis, a scalable data curation and
distillation pipeline for transferring
multimodal reasoning capabilities from high-capacity teacher models to smaller, deployment-oriented MLLMs. Starting from a diverse open-source seed pool, our pipeline generates structured CoT traces and performs joint annotation of
reasoning difficulty, answer quality, and semantic task tags. To maintain data quality at scale, we combine rule-based filtering, difficulty-aware selection, and tag-based diversity sampling, resulting in a curated corpus of 1.8M samples that supports controllable subset construction for downstream training. We use OmniThoughtVis to distill Qwen3-VL models from 2B to 8B parameters and evaluate them on nine
multimodal reasoning benchmarks. The resulting distilled models show consistent gains across model scales, including improvements of up to +16.8 points on MathVerse and +5.6 points on MMMU-Pro for the 4B model. Notably, the distilled 4B model matches or surpasses the undistilled 8B baseline on several tasks, highlighting the practical value of scalable
reasoning distillation for deployment-oriented MLLMs.
Read the full abstract ▾
All authors Yuanhao Yue, Chengyu Wang, Yuanjie Lyu, Lei Shen, Jun Huang