论文检索

Applications · Language, Speech and Dialog

Qiuchen Wang, Shihang Wang, Yu Zeng, Qiang Zhang, Fanrui Zhang, Zhuoning Guo, Bosi Zhang, Wenxuan Huang, Lin Chen, Zehui Chen 等

Effectively retrieving, reasoning, and understanding multimodal information remains a critical challenge for agentic systems. Traditional Retrieval-augmented Generation (RAG) methods rely on linear interaction histories, which struggle to handle long-context tasks, especially those involving information-sparse yet token-heavy visual data in iterative reasoning scenarios. To bridge this gap, we introduce VimRAG, a framework tailored for multimodal Retrieval-augmented Reasoning across text, images, and videos. Inspired by our systematic study, we model the reasoning process as a dynamic directed acyclic graph that structures the agent states and retrieved multimodal evidence. Building upon this structured memory, we introduce a Graph-Modulated Visual Memory Encoding mechanism, with which the significance of memory nodes is evaluated via their topological position, allowing the model to dynamically allocate high-resolution tokens to pivotal evidence while compressing or discarding trivial clues. To implement this paradigm, we propose a Graph-Guided Policy Optimization strategy. This strategy disentangles step-wise validity from trajectory-level rewards by pruning memory nodes associated with redundant actions, thereby facilitating fine-grained credit assignment. Extensive experiments demonstrate that VimRAG consistently achieves state-of-the-art performance on diverse multimodal RAG benchmarks.

Applications · Language, Speech and Dialog

Doyoung Kim, Youngjun Lee, Joeun Kim, Jihwan Bang, Hwanjun Song, Susik Yoon, Jae-Gil Lee

Conversational query reformulation (CQR) has become indispensable for improving retrieval in dialogue-based applications. However, existing approaches typically rely on reference passages for optimization, which are **impractical** to acquire in real-world scenarios. To address this limitation, we introduce a novel **reference-free** preference optimization framework ***DualReform*** that generates **pseudo reference passages** from **commonly-encountered** conversational datasets containing only queries and responses. DualReform attains this goal through two key innovations: (1) **response-based inference**, where responses serve as proxies to infer pseudo reference passages, and (2) **response refinement via the dual-role of CQR**, where a CQR model refines responses based on the shared objectives between response refinement and CQR. Despite not relying on reference passages, ***DualReform*** achieves 96.9--99.1% of the retrieval accuracy attainable only with reference passages and surpasses the state-of-the-art method by up to 31.6%.

Deep Learning · Large Language Models

Ben Rank, Hardik Bhatnagar, Ameya Pandurang Prabhu, Shira Eisenberg, Karina Nguyen, Matthias Bethge, Maksym Andriushchenko

Given the recent rapid progress of LLM agents like Claude Code or Codex CLI for software engineering, an important next question is whether they can automate AI research itself. In this paper, we study *post-training*, which is the critical step that turns base LLMs into useful assistants. We introduce PostTrainBench to benchmark how well LLM agents can perform post-training *autonomously* under bounded compute constraints (10 hours on one H100 GPU). We task frontier agents (e.g., Claude Code with Opus 4.5) to optimize the performance of a base LLM on a particular benchmark (e.g., Qwen3-4B on AIME). Importantly, we do not provide any predefined strategies to the agents and instead give them full autonomy to find necessary information on the web, run experiments, and curate data. We find that frontier agents make substantial progress but generally lag behind instruction-tuned LLMs from leading providers: 21.5% for the best agent vs. 51.1% for official instruction-tuned models. However, agents can exceed instruction-tuned models in targeted scenarios: GPT-5.1 Codex Max achieves 89% on BFCL with Gemma-3-4B vs. 67% for the official model. Additionally, we document concerning behaviors related to reward hacking, such as training on test data or downloading pre-existing instruction-tuned models, and unauthorized usage of API keys for synthetic data generation. Overall, we expect PostTrainBench to serve as an important benchmark for tracking both capabilities and risks of AI R&D automation.

Deep Learning · Large Language Models

Yuan Li, Bo Wang, Yufei Gao, Yuqian Yao, Xinyuan Wang, Zhangyue Yin, Xipeng Qiu

Proximal constraints are fundamental to the stability of the Large Language Model reinforcement learning. While the canonical clipping mechanism in PPO serves as an efficient surrogate for trust regions, we identify a critical bottleneck: fixed bounds strictly constrain the upward update margin of low-probability actions, disproportionately suppressing high-advantage tail strategies and inducing rapid entropy collapse. To address this, we introduce **Band-constrained Policy Optimization** (BandPO). BandPO replaces canonical clipping with **Band**, a unified theoretical operator that projects trust regions defined by $f$-divergences into dynamic, probability-aware clipping intervals. Theoretical analysis confirms that Band effectively resolves this exploration bottleneck. We formulate this mapping as a convex optimization problem, guaranteeing a globally optimal numerical solution while deriving closed-form solutions for specific divergences. Extensive experiments across diverse models and datasets demonstrate that BandPO consistently outperforms canonical clipping and Clip-Higher, while robustly mitigating entropy collapse.

Deep Learning · Large Language Models

Heming Zou, Yixiu Mao, Yun Qu, Qi Wang, Xiangyang Ji

Supervised fine-tuning (SFT) is a commonly used technique to adapt large language models (LLMs) to downstream tasks. In practice, SFT on a full dataset is computationally expensive and sometimes suffers from overfitting or bias amplification. This facilitates the rise of data curation in SFT, which prioritizes the most valuable data to optimze. This work studies the online batch selection family that dynamically scores and filters samples during the training process. However, existing popular methods often (i) rely merely on the utility of data to select a subset while neglecting other crucial factors like diversity, (ii) rely on external resources such as reference models or validation sets, and (iii) incur extra training time over full-dataset training. To address these limitations, this work develops UDS (Utility-Diversity Sampling), a framework for efficient online batch selection in SFT. UDS leverages the nuclear norm of the logits matrix to capture both data utility and intra-sample diversity, while estimating inter-sample diversity through efficient low-dimensional embedding comparisons with a lightweight memory buffer of historical samples. Such a design eliminates the need for external resources and unnecessary backpropagation, securing computational efficiency. Experiments on multiple benchmarks demonstrate that UDS consistently outperforms state-of-the-art online batch selection methods under varying data budgets, and significantly reduces training time compared to full-dataset fine-tuning.

Deep Learning · Large Language Models

Tianyi Wang, Long Li, Hongcan Guo, Yibiao Chen, Yixia Li, Yong Wang, Yun Chen, Guanhua CHEN

Reinforcement Learning with Verifiable Rewards (RLVR) is increasingly viewed as a tree pruning mechanism. However, we identify a systemic pathology termed Recursive Space Contraction (RSC), an irreversible collapse driven by the combined dynamics of positive sharpening and negative squeezing, where the sampling probability of valid alternatives vanishes. While Kullback-Leibler (KL) regularization aims to mitigate this, it imposes a rigid Shape Matching constraint that forces the policy to mimic the reference model's full density, creating a gradient conflict with the sharpening required for correctness. We propose Anchored Policy Optimization (APO), shifting the paradigm from global Shape Matching to Support Coverage. By defining a Safe Manifold based on the reference model's high-confidence support, APO permits aggressive sharpening for efficiency while selectively invoking a restorative force during error correction to prevent collapse. We theoretically derive that APO serves as a gradient-aligned mechanism to maximize support coverage, enabling an Elastic Recovery that re-inflates valid branches. Empirical evaluations on mathematical benchmarks demonstrate that APO breaks the accuracy-diversity trade-off, significantly improving Pass@1 while restoring the Pass@K diversity typically lost by standard policy gradient methods.

Applications · Neuroscience, Cognitive Science

Connor Lane, Ratna Grandhi, Leema Krishna Murali, Mihir Tripathy, Shamus Zi Yang Sim, Will Beddow, Gianfranco Cortes, Suin Cho, Debojyoti Das, Sam Gijsen 等

We propose a simple strategy for training a foundation model on functional MRI (fMRI) data: we adapt the standard Vision Transformer to fMRI by first converting each 3D fMRI volume to a 2D map using a standard cortical flat map projection. We train spatiotemporal masked autoencoders (MAE) on 2.3K hours of fMRI flat map videos. Our model (CortexMAE) outperforms identical MAE models trained on parcel-averaged or native volume data. We perform the first quantitative scaling analyses for fMRI and observe strict power law scaling. Finally, we develop the first open evaluation suite for fMRI foundation models and use it to perform a comprehensive comparison. On cognitive state decoding, our model outperforms all models by a wide margin. On clinical trait prediction, however, we report an important mixed result: all models show inconsistent performance (including our own). We hope that by introducing reproducible benchmarks and a strong, simple baseline, we can help establish a clear frontier for fMRI foundation models. Code is available at \url{https://anonymous.4open.science/r/cortex_mae}.

Deep Learning · Large Language Models

Yingpeng Ma, Jianhao Yan, Bei Shi, Ka Hou Kam, Runnan Wang, Xuebo Liu, Yulong Chen, Yue Zhang, Derek F. Wong

The rapid advancement of Large Language Models (LLMs) is revolutionizing AI for Game by enabling open-ended and fluid interactive storytelling. However, existing research has largely overlooked the critical challenge of maintaining logical consistency and narrative integrity against unconstrained user interventions. To address this, we formulate this challenge as \emph{Narrative Commitment Preservation (NCP)}, and take interactive narrative as our testbed. We introduce NCP-Bench, a benchmark of 100 narrative environments derived from movie synopses. Each environment includes a structured narrative specification (trajectory, commitments, and initial facts) that we can reliably check throughout the interaction between player and storyteller. Experiments across state-of-the-art LLMs reveal that high linguistic quality does not guarantee commitment preservation, even strong models frequently generate logically conflicting content under adversarial interventions, with the best-performing model (GPT-5.2) achieving only 40\% survival rate after 20 turns and fact conflicts occurring in 40\%--68\% of all interactions.

Deep Learning · Large Language Models

Xuan Shen, Yizhou Wang, Yufa Zhou, Xiangxi Shi, Pu Zhao, Yanzhi Wang, Jiuxiang Gu

Chain-of-Thought (CoT) reasoning has become a powerful framework for improving complex problem-solving capabilities in Multimodal Large Language Models (MLLMs). However, the verbose nature of textual reasoning introduces significant inefficiencies. In this work, we propose**Heima** (as hidden llama), an effective CoT compression framework that condenses lengthy CoTs into a small set of abstract thinking tokens, preserving essential reasoning while removing redundancy. We then conduct a theoretical analysis from an information-theoretic perspective, quantifying the information gap induced by compression, showing that reasoning capability is preserved when non-trivial mutual information is retained. To further explore and quantify this information gap, we design the adaptive interpreter that maps thinking tokens back to variable-length textual sequences, thereby reconstructing the reasoning process. Experiments across diverse reasoning benchmarks demonstrate that Heima improves reasoning efficiency, while maintaining or even achieving better zero-shot accuracy. Moreover, the interpreter reconstructs coherent reasoning progresses from compressed thinking tokens, revealing that the information gap is minimal and validating the effectiveness of the proposed framework. This work paves the way for scalable latent reasoning models and advances our understanding of efficient reasoning processes in large models.

Theory · Game Theory

Haris Aziz, Ling Gai, Yuhang Guo, Jeremy Vollen

We study the transit stop placement (TrSP) problem in general metric spaces, where agents travel between source–destination pairs and may either walk directly or utilize a shuttle service via selected transit stops. We investigate fairness in TrSP through the lens of justified representation (JR) and the core, and uncover a structural correspondence with fair clustering. Specifically, we show that a constant-factor approximation to proportional fairness in clustering can be used to guarantee a constant-factor bi-parameterized approximation to core. We establish a lower bound of $1.366$ on the approximability of JR, and moreover show that no clustering algorithm can approximate JR within a factor better than $3$. Going beyond clustering, we propose the Expanding Cost Algorithm, which achieves a tight $2.414$-approximation for JR, but does not give any bounded core guarantee. In light of this, we introduce a parameterized algorithm that interpolates between these approaches, and enables a tunable trade-off between JR and core. Finally, we complement our results with an experimental analysis using small-market public carpooling data.

Deep Learning · Large Language Models

Mingzi Wang, Lancheng Zou, Shuo Yin, Zhuolun He, Bei Yu

The progressive scaling of large language models (LLMs) has consistently enhanced multimodal understanding and advanced reasoning capabilities, but has substantially increased computational and hardware execution overhead. In this paper, we present S-Quant, a novel post-method that compresses only model weights. We partition each weight tensor into fixed-size blocks and assign a single seed to each block. The seed drives a hardware-friendly Linear Feedback Shift Register (LFSR) generator that dynamically produces multiple basis matrices. Each block is then reconstructed as a linear combination of these basis matrices, with block-specific coefficients, which substantially reduces the amount of stored data, increases the data-transfer efficiency between memory and compute units, and consequently speeds up memory-bound inference for large language models. Experimental results on different LLM models ranging from 7B–70B parameters show that S-Quant attains state-of-the-art performance when weights are compressed to approximately 3-bit or 4-bit. We also design a dedicated ASIC accelerator that achieves a 4× speed-up for memory-bound LLM inference.

Deep Learning · Large Language Models

Xiaojun Guo, Runyu Zhou, Yifei Wang, Qi Zhang, Chenheng Zhang, Stefanie Jegelka, Xiaohan Wang, Jiajun Chai, Guojun Yin, Wei Lin 等

Vision-language models (VLMs) have shown remarkable abilities by integrating large language models with visual inputs. However, they often fail to utilize visual evidence adequately, either depending on linguistic priors in vision-centric tasks or resorting to textual shortcuts during reasoning. Although reinforcement learning (RL) can align models with desired behaviors, its application to VLMs has been hindered by the lack of scalable and reliable reward mechanisms. To overcome this challenge, we propose **SSL4RL**, a novel framework that leverages self-supervised learning (SSL) tasks as a source of verifiable rewards for RL-based fine-tuning. Our approach reformulates SSL objectives—such as predicting image rotation or reconstructing masked patches—into dense, automatic reward signals, eliminating the need for human preference data or unreliable AI evaluators. Experiments show that SSL4RL substantially improves performance on both vision-centric and vision-language reasoning benchmarks, with encouraging potentials on open-ended image-captioning scenarios and stronger resilience to visual corruptions. Through systematic ablations, we identify key factors—such as data volume, model scale, model choice, task difficulty, and semantic alignment with the target domain—that influence the effectiveness of SSL4RL tasks, offering new design principles for future work. We also demonstrate the framework’s generality by applying it to graph learning, where it yields significant gains. SSL4RL establishes a versatile and effective paradigm for aligning multimodal models using verifiable, self-supervised objectives.

Deep Learning · Large Language Models

Minsoo Kim, Arnav Kundu, Han-Byul Kim, Richa Dixit, Minsik Cho

Modern large language models (LLMs) extend context lengths to millions of tokens, enabling coherent, personalized responses grounded in long conversational history. However, the Key-Value (KV) cache grows linearly with the extended dialogue history, causing the model’s memory footprint to quickly exceed device limits. While recent KV cache compression methods attempt to reduce memory usage, most apply cache eviction after processing the entire context, incurring unbounded peak memory usage. Additionally, query-dependent eviction narrows the cache semantics to a single query, leading to failure cases in multi-turn conversations. In this paper, we introduce EpiCache, a training-free KV cache management framework for long conversational question answering (LongConvQA) under fixed memory budgets. EpiCache bounds cache growth through block-wise prefill and preserves topic-relevant context via episodic KV compression, which clusters conversation history into coherent episodes and performs episode-specific KV cache eviction. Across three LongConvQA benchmarks (LongMemEval, Realtalk, and LoCoMo), EpiCache improves accuracy by up to 30\%, achieves near-full-cache accuracy under $4$–$6\times$ compression, and reduces latency and peak memory by up to $2.4\times$ and $3.7\times$, respectively.

Deep Learning · Graph Neural Networks

Ayushman Raghuvanshi, Thummaluru Siddartha Reddy, Sundeep Prabhakar Chepuri, Mahesh Chandran

Continuous-time dynamic graphs (CTDGs) provide a richer framework to capture fine-grained temporal patterns in evolving relational data. Long-range information propagation is a key challenge while learning representations, wherein it is important to retain and update information over long temporal horizons. Existing approaches restrict models to capture one-hop or local temporal neighborhoods and fail to capture multi-hop or global structural patterns. To mitigate this, we derive a parameter-efficient state-space modeling framework for continuous-time dynamic graphs $\texttt{(CTDG-SSM)}$ from first principles. We first introduce continuous-time Topology-Aware higher order polynomial projection operator ($\texttt{CTT-HiPPO}$), a novel memory-based reformulation of $\texttt{HiPPO}$ to jointly encode temporal dynamics and graph structure. The solution from $\texttt{CTT-HiPPO}$ are obtained by projecting the classical HiPPO solution through a polynomial of the Laplacian matrix, yielding topology-aware memory updates that admit an equivalent state-space formulation for CTDGs ($\texttt{CTDG-SSM}$). Then a computationally efficient discrete formulation is obtained using the zero-order hold approach for model implementation. Across benchmarks on dynamic link prediction, dynamic node classification, and sequence classification, $\texttt{CTDG-SSM}$ achieves state-of-the-art performance. Notably, it achieves large performance gains on datasets that require long range temporal (LRT) and spatial reasoning.

Deep Learning · Large Language Models

Wenhao Li, Jinhao Dong, Hailin Zhang, Wenhang Shi, WEI LU, Xiaoyong Du

Long-context Large Language Model inference is severely bottlenecked by the massive Key-Value (KV) cache, yet existing sparse attention methods often suffer from static fixed-budget (Top-k) retrieval or rely on proxy scores that are computationally expensive and biased. To address these limitations, we propose RaBitQCache, a novel sparse attention framework that utilizes randomized rotated binary quantization and high-throughput binary-INT4 arithmetic to efficiently estimate attention weights. Our proxy score serves as an unbiased estimator with a proven error bound, enabling adaptive Top-p retrieval that dynamically adjusts the token budget based on actual attention sparsity. We further implement a hardware-aware system with asynchronous pipelining and lazy updates to mask overhead. Evaluations demonstrate that RaBitQCache significantly accelerates inference and reduces memory I/O while preserving generation quality compared to state-of-the-art baselines.

Deep Learning · Foundation Models

Wenhan Ma, Hailin Zhang, Liang Zhao, Yifan Song, Yudong Wang, Fuli Luo, Zhifang Sui

Reinforcement learning (RL) has emerged as a crucial approach for enhancing the capabilities of large language models. However, in Mixture-of-Experts (MoE) models, the routing mechanism often introduces instability, even leading to catastrophic RL training collapse. We analyze the training-inference consistency of MoE models and identify a notable discrepancy in routing behaviors between the two phases. To address this issue, we propose \textbf{Rollout Routing Replay (R3)}, a novel and effective method that records routing distributions from the inference engine and replays them during training. R3 significantly reduces training-inference policy KL divergence and mitigates extreme discrepancies without compromising training speed. Extensive experiments on various settings confirm that R3 succeeds in stabilizing RL training, preventing collapse and outperforming strong baselines. R3 is orthogonal to most policy optimization algorithm improvements, allowing it to be used in conjunction with them. We believe this work can offer a new solution for stabilizing RL in MoE model.

Applications · Neuroscience, Cognitive Science

Yamin Li, Shiyu Wang, Chang Li, Ange Lou, Haatef Pourmotabbed, Sarah Goodale, Dario Englot, Daniel Moyer, Roza G Bayrak, Catie Chang

Functional magnetic resonance imaging (fMRI) provides dynamic measurements of human brain activity at high spatial resolution and depth, but its use is constrained by high cost, limited accessibility, and strict acquisition requirements. Synthesizing fMRI data from more accessible, non-invasive modalities such as electroencephalography (EEG) offers a promising alternative, enabling inference of deep brain dynamics from low-cost scalp recordings in naturalistic settings. Despite recent progress, existing EEG-to-fMRI translation methods typically rely on region-specific models and offer limited support for subject-level and dataset-level heterogeneity, restricting their generalizability. We propose UniEFS, a unified EEG-to-fMRI synthesis model that enables full-brain fMRI reconstruction while accommodating varying demographic and physiological contexts within a single model. Our approach leverages a pretrained fMRI decoder to embed rich spatial priors and introduces condition-aware prompt tokens that encode subject-level and experimental metadata, enabling effective handling of heterogeneous datasets. We extensively evaluate our model performance on eyes-closed resting-state data and demonstrate that it can reliably reconstruct temporally-resolved whole-brain fMRI activity, with strong potential to generalize to task-based fMRI and clinical populations in a zero-shot manner.

Deep Learning · Attention Mechanisms

Xiaodong Ji, Hailin Zhang, Fangcheng Fu, Bin Cui

Many advanced Large Language Model (LLM) applications require long-context processing, but the self-attention module becomes a bottleneck during the prefilling stage of inference due to its quadratic time complexity with respect to sequence length. Existing sparse attention methods accelerate attention computation by skipping less significant regions of the attention map. However, these approaches typically perform coarse-grained inspection of the attention map, resulting in their suboptimal performance. In this paper, we propose SALE, a fine-grained sparse attention method that accelerates the long-context prefilling stage of LLM with negligible loss in model accuracy. SALE achieves fast and accurate fine-grained attention map estimation using low-bit quantized query-key products to approximate attention weights, followed by the application of a novel Relative Attention Score metric to assess the importance of query-key pairs. This design enables us to accurately identify important regions in the attention map, thereby constructing a highly sparse attention mask. We implement a custom CUDA kernel in SALE optimized for hardware efficiency, reducing overhead to approximately 11% of the full attention latency. Notably, SALE requires no parameter training and can be seamlessly integrated into existing systems with trivial code modifications. Experiments on long-context benchmarks demonstrate that our method outperforms existing approaches in accuracy-efficiency trade-offs, achieving at least 3.36× speedups on Llama-3.1-8B for sequences longer than 64K while maintaining model quality.

Reinforcement Learning · Everything Else

Wei Chen, Yubing Wu, Junmei Yang, Delu Zeng, Qibin Zhao, John Paisley, Min Chen, Zhou Wang

Preference optimization is widely used to align large language models (LLMs) with human preferences, yet many margin-based objectives often suppress the chosen response together with the rejected one, and no general mechanism exists to prevent this across objectives. We bridge this gap by presenting a unified \textbf{incentive-score decomposition} of preference optimization, revealing that diverse objectives share identical local update directions and differ only in their scalar weighting coefficients. Building on this decomposition, by analyzing the dynamics of the rewards of chosen/rejected responses, we identify the \textbf{disentanglement band (DB)}, a simple, testable condition that characterizes when training can realize the ideal pathway: suppressing the loser while maintaining the winner, possibly after an initial transient. Leveraging the DB, we propose a plug-and-play \textbf{reward calibration (RC)} that adaptively rebalances chosen versus rejected updates to satisfy the DB, without redesigning the base objective. Empirical results confirm that this calibration effectively disentangles updates and improves alignment performance across diverse objectives.

Applications · Robotics

Haoming Xu, Lei Lei, Jie Gu, Chu Tang, Jingmin Chen, Rui-Qi Wang

We present Move-Then-Operate, a Vision–language–action framework that explicitly decouples robotic manipulation into two distinct behavioral phases: coarse relocation (move) and contact-critical interaction (operate). Unlike monolithic policies that conflate these heterogeneous regimes, our architecture employs a dual-expert policy routed by a learnable phase selector, introducing a structural inductive bias that isolates phase-specific dynamics. Phase labels are automatically generated via an MLLM-based pipeline conditioned on lightweight contextual cues such as end-effector velocity and subtask decomposition to ensure alignment with human motor patterns. Evaluated on the RoboTwin2 benchmark, our method achieves an average success rate of $68.9\%$, outperforming the monolithic $\pi_0$ baseline by +$24\%$. It matches or exceeds models trained on $10\times$ more data and reaches peak performance in $40\%$ fewer training steps, demonstrating that architectural disentanglement of move and operate phases is a highly effective and efficient strategy for mastering high-precision manipulation.