Searcharxiv⌕ Search

arXiv subjects

Xinran Zhang

Publications and source records attributed to Xinran Zhang.

At least 19 recordsLinked to original sources

OpticalRAG: Pixel-Space Compression for Token-Efficient Retrieval-Augmented Generation

Retrieval-augmented generation (RAG) is bottlenecked by the token budget of large language models: feeding long documents into the context window is expensive, while aggressive textual compression discards fine-grained information that is irreversibly lost in a one-dimensional token sequence. We argue that this bottleneck is not fundamental to information density but to the choice of compression \emph{modality}. Modern visual encoders can pack the contents of a rendered text page into roughly a hundred visual tokens with little semantic loss, opening a new compression-fidelity tradeoff that purely textual methods cannot reach. We present \textbf{OpticalRAG}, a framework that renders documents as page images, encodes them with a frozen vision encoder, and distills the retrieved visual tokens into as few as $16$ injected tokens before passing them to the LLM. Two ingredients make this work without expensive cross-modal training: (i) \emph{Encoder-Bridged Retrieval}, which projects text-query states into the visual feature space of the frozen encoder, through a projection shared with the compressor, to retrieve visual tokens from a text query, with no contrastive training; and (ii) \emph{Query-Driven Distillation}, a lightweight cross-attention compressor that condenses the retrieved visual tokens into a small set of query-conditioned prefix tokens. On LongBench-v2, BAMBOO, and LooGLE-v2, OpticalRAG achieves the best compression-aware accuracy (CAP) on all three benchmarks while injecting only $16$ tokens, and improves the closed-book Qwen2.5-7B-Instruct backbone by $5.17$ points on LongBench-v2. Code is available at https://github.com/brighterr/OpticalRAG.

cs.CL↗

Breathing mode of quantum droplets in dipolar quantum gases: A sum-rule analysis

We investigate theoretically the ground-state properties and breathing-mode collective excitations of three-dimensional dipolar Bose gases in anisotropic harmonic traps incorporating quantum fluctuations. Combining a Gaussian variational ansatz with a sum-rule analysis, we derive explicit analytical expressions for both axial and radial breathing-mode frequencies, which are validated by numerical solutions of the time-dependent extended Gross-Pitaevskii equation. Our theoretical predictions show great agreement with existing experimental data for $^{166}$Er and $^{162}$Dy gases. By constructing comprehensive phase diagrams across the parameter space of the $s$-wave scattering length, atom number, and trap aspect ratio, we reveal both discontinuous first-order phase transitions and smooth crossovers between the dilute Bose-Einstein condensate and dense quantum droplet phases. We confirm that the enhanced incompressibility induced by quantum fluctuations significantly elevates the breathing-mode frequencies in the droplet phase compared with conventional weakly interacting Bose gases. Furthermore, the system undergoes a phase transition and a crossover over the scattering length under the elongated and flattened confinements, characterized by discontinuous jumps and continuous crossovers in peak density and atomic cloud sizes, respectively. Our work offers a useful analytical framework to characterize collective excitations in dipolar quantum gases, providing quantitative insights for forthcoming ultracold atom experiments in lanthanide atoms and polar molecules.

cond-mat.quant-gas↗

CTE-Bench: Counterfactual Trace Evaluation for Stateful Software Simulators

Coding agents change running software: they patch a service's code or overwrite its stored state, and then act on their own expectation of how the service will respond afterwards. A wrong expectation may surface only several calls later. Function-level code-execution benchmarks omit persistent service state, and agent benchmarks score the actions an agent takes or the final state it reaches. We introduce CTE-Bench, which measures whether a model can predict how an intervention changes a stateful service's future behavior, without asking it to choose actions. Each scenario gives the model Python service code, the calls and responses observed before the intervention, the intervention itself (a source edit or a state overwrite), and 40 fixed future calls; the model predicts every future response, and predictions are checked by executing the service. Three memory protocols control whether the model sees the correct earlier responses, none of them, or its own earlier predictions. CTE-Bench-Core-v1 contains 255 scenarios over six deterministic Python services, giving 10,200 predictions per model. The main score is effect-step value match (VM): exact response equality on the 2,476 future calls whose response the intervention changes. With correct earlier responses revealed, four API-hosted models (DeepSeek V4-Flash, Kimi K2.5, Qwen3.6-35B-A3B, and Claude Sonnet 4.6) reach 54.3%-61.7% effect-step VM. Hiding those responses lowers effect-step VM to 23.2%-29.0%; conditioning on self-generated predictions gives 24.8%-33.2%, and at most 1.2% of scenarios are predicted exactly end to end. Current models thus track intervention effects mainly when correct feedback is supplied, and their errors compound over a rollout. We release CTE-Bench-Core-v1 with its executable oracle, evaluation scripts, and an evaluation card mapping each claim to its protocol.

cs.SE↗

Towards Generalist Game Agents: A Survey of Foundation Models in Virtual Worlds

The real world unfolds along a single set of physics laws, yet human intelligence demonstrates a remarkable capacity to generalize experiences from this singular physical existence into a multiverse of games, each governed by entirely different rules, aesthetics, physics, and objectives. This omni-reality adaptability is a hallmark of general intelligence. As Artificial Intelligence progresses towards Artificial General Intelligence, the multiverse of games has evolved from mere entertainment into the ultimate ground for training and evaluating AGI. The pursuit of this generality has unfolded across four eras: from environment-specific symbolic and reinforcement learning agents, to current large foundation models acting as generalist players, and toward a future creator stage where agent both creates new game worlds and continually evolves within them. We trace the full lifecycle of a generalist game player along four interdependent pillars: Dataset, Model, Harness, and Benchmark. Every advance across these pillars can be read as an attempt to break one of five fundamental trade-offs that currently bound the whole system. Building on this end-to-end view, we chart a five-step roadmap, progressing from single-game mastery to the ultimate creator stage in which the agent simultaneously creates and evolves within theoretical game multiverse. Taken together, our work offers a unified lens onto a rapidly shifting field, and a principled path toward the omnipotent generalist agent capable of seamlessly mastering any challenge within the multiverse of games, thereby paving the way for AGI.

cs.CV↗

Agents' Last Exam

Recent AI systems have achieved strong results on a wide range of benchmarks, yet these gains have not translated into economically meaningful deployment across many professional domains. We argue that this gap is largely an evaluation problem: widely used benchmarks lack sustained performance measurement on real and economically valuable workflows. This paper introduces Agents' Last Exam (ALE), a benchmark designed to evaluate AI agents on long horizon, economically valuable, real world tasks with verifiable outcomes. Developed in collaboration with 250+ industry experts, ALE covers non-physical industries defined with reference to O*NET / SOC 2018 (the U.S. federal occupational taxonomy). It is organized around a task taxonomy with 55 sub fields grouped into 13 industry clusters covering 1K+ tasks. Current results show that the hardest tier remains far from saturated: across mainstream harness and backbone configurations, the average full pass rate is below 1%. ALE is designed as a living benchmark: its task pool grows continuously as new workflows and industries are onboarded. More broadly, ALE is intended not merely as another leaderboard, but as an instrument for closing the gap between benchmark success and GDP relevant impact.

cs.AI↗

Atomic and Holistic LLM Judges for Reference-Grounded Support Labels: A Prompt-Controlled Comparison

When an LLM judge only has to assign a three-way support label to a candidate answer given a reference, does asking it to decompose the answer into atomic claims help, and at what cost? We compare four single-call designs that share the judge model, the inputs, and the level of instruction detail: candidate-side atomic decomposition, a matched holistic rubric, reference-side decomposition that checks whether each reference claim is covered, and a bidirectional combination. Support labels are constructed from TruthfulQA, ASQA, and QAMPARI references (200 questions and 400 rows per dataset). All four designs are run with Opus-4.6, GPT-4.1, and Gemini Flash Lite; Sonnet-4.6 is added for the candidate-side and holistic designs. Candidate-side decomposition is weak where the label depends on completeness: the holistic rubric is more accurate on ASQA and QAMPARI for every judge while using fewer tokens. Reference-side decomposition is 12.5-21.3 points more accurate than holistic on ASQA for 28-29% more tokens, and stays near the holistic ceiling on QAMPARI at 63-69% more. On TruthfulQA misconceptions, candidate-side decomposition is competitive and significantly better for two judges. On a 60-row single-author subset whose labels are looser than strict reference completeness, candidate-side decomposition edges ahead of holistic for every judge, reversing the construction-label order. What a judge decomposes should follow what the label measures.

cs.CL↗

TACO: TActile World Model as a Self-COrrector for Scalable Robot Policy Post-Training

Vision-Language-Action models and World Action Models have shown promising generalization in robotic manipulation but remain fragile in contact-rich tasks, where contact perturbations can cause failures that are difficult to detect from vision alone. Corrective post-training with tactile feedback can improve recovery, but scaling such supervision through human intervention is costly. World models can synthesize additional training data, yet vision-only generation may produce visually plausible but contact-inconsistent trajectories. We therefore introduce TACO, a scalable robot policy post-training framework built on a compositional tactile world model. Given real rollouts, TACO follows a Recognize--Imagine--Label loop: an inverse dynamics and value model identifies failure-adjacent states using progress estimates, a visuo-tactile generation model imagines local corrections by jointly generating video and tactile sequences, and the inverse dynamics and value model labels them with corrective actions and progress scores. Candidates are filtered for kinematic feasibility and tactile plausibility, then selected by predicted progress gain. TACO aggregates demonstrations, real rollouts, and selected corrections for iterative post-training. It combines knowledge-insulated tactile adaptation with CFG-RL using binary advantage labels while keeping the pretrained VLM backbone fixed. Experiments on real-world tasks show that TACO improves the average task score from 0.375 to 0.825 after two post-training iterations.

cs.RO↗

ERPBench: Evaluating LLM Agents for Enterprise Decision-Making Across Competitive Market Ecologies

Large language model (LLM) agents are increasingly proposed for enterprise workflows, yet existing evaluations rarely test whether business-decision conclusions transfer across competitive market ecologies. We introduce ERPBench, an execution-instrumented benchmark for enterprise decision agents in a six-round Enterprise Resource Planning (ERP) simulation with coupled pricing, production, procurement, inventory, finance, and shared-market competition. ERPBench evaluates the same 100 fixed problems in two matched competitive market ecologies: Solo, where each evaluated LLM agent competes against fixed rule-based opponents, and Arena, where six evaluated LLM agents compete in a shared market. Across six model families, this yields 1,200 model-level trajectories spanning 7,200 decision rounds. Under the observed service configuration, the leading model differs between ecologies: DeepSeek leads in Solo (252.29M mean valuation; mean rank 1.67), whereas Gemini leads in Arena (263.95M; 1.76). The two ecologies identify the same task-level winner on only 21 of 100 problems, and Gemini's bottom-rank rate falls from 22 % to 0 % in Arena. ERPBench supports paired evaluation of whether enterprise-agent rankings transfer across competitive market ecologies, supplemented by aggregate execution-intervention analysis. Code and benchmark resources are available in our https://github.com/GAIR-NLP/erp-bench.

cs.AI↗

Towards Spatial Supersensing in the Wild

Humans can efficiently parse continuous sensory streams, from hours to years, scaffolding an internal world model that grounds spatial reasoning and prediction. To mimic this capacity, spatial supersensing challenges multimodal models to move beyond linguistic understanding toward true world modeling. However, their benchmark relies on synthetic long videos, formed by concatenating random short clips, and is mostly limited to household scenes, leaving real-world continuity and diversity underexplored. To address the gap, we introduce $\textbf{VSI-Super-Wild}$, a large-scale benchmark for evaluating spatial supersensing over long temporal horizons in diverse in-the-wild scenes. Notably, inspired by cognitive studies on how humans structure experience, we systematically probe the full triad of world state: the agent (observer), objects (scene items), and the environment (places and global layout). In total, VSI-Super-Wild contains $\textbf{6,980}$ human-verified question-answer pairs derived from $\textbf{442}$ real-world videos spanning 8 scene categories, including long-form recordings exceeding 4 hours. Results on VSI-Super-Wild expose a fundamental disconnect: despite advances in static image understanding, models consistently fail at tasks that require coherent world-state tracking over time. We characterize how performance degrades with world-state complexity and temporal horizon, and diagnose four failure modes: spatial collapse, semantic shortcuts, insufficient update, and instance confusion. This taxonomy reveals that models lack mechanisms to bind objects, agents, and environments into a unified spatial world model, a fundamental gap that defines the path forward for spatial supersensing.

cs.CV↗

One-dimensional asymmetrically interacting quantum droplets in Bose-Bose mixtures

We theoretically investigate ground-state properties and collective excitations of one-dimensional quantum droplets in asymmetric Bose-Bose mixtures with unequal intraspin interactions. Using the extended Gross-Pitaevskii equation supported by variational, sum-rule, and linearization methods, we show that the intraspin interaction ratio substantially alters the droplet's density profile, driving a transition from Gaussian-like to flat-top shapes. By examining two experimentally relevant parameter regions, we analyze density profiles, radii, peak densities, and excitation spectra to distinguish quantum phases and to depict phase diagrams in the space of asymmetric interaction ratio and total atom number. We carefully study the frequencies of both well-known dipole and breathing modes and less-explored spin-dipole and spin-breathing modes. The breathing-mode frequency decreases monotonically with interaction ratio, approaching asymptotically the result of a conventional weakly interacting Bose gas. It varies nonmonotonically with total atom number, peaking at a critical point that highlights the crucial role of quantum fluctuations. In contrast, spin modes display distinct temporal spin density distributions and reveal in-phase and out-of-phase relative dynamics between components. Their frequencies depend instead monotonically on the interaction ratio and atom number. Our results provide a comprehensive understanding of asymmetric quantum droplets and link to experimentally accessible regimes in ultracold $^{39}$K atomic gases.

cond-mat.quant-gas↗

Distributed Generative Inference of LLM at Internet Scales with Multi-Dimensional Communication Optimization

Decentralized LLM inference distributes computation among heterogeneous nodes across the internet, offering a performant and cost-efficient solution, alternative to traditional centralized inference. However, the low cross-node network bandwidth makes communication the primary bottleneck. In this paper, we introduce BloomBee, an internet-scale distributed LLM inference framework. BloomBee integrates LLM-layer assignment, micro-batching and tensor offloading to optimize communication from multiple dimensions. Additionally, BloomBee formulates the coordination of these techniques as an optimization problem and solves it using dynamic programming. BloomBee also customizes lossless compression and speculative decoding according to low-bandwidth network settings to reduce communication overhead. We evaluate BloomBee across a spectrum of network environments and show that it improves service throughput by up to 1.76x. It also reduces average latency by up to 43.20% compared to state-of-the-art decentralized LLM inference systems. BloomBee is open-sourced.

cs.DC↗

How Sensitive Are Safety Benchmarks to Judge Configuration Choices?

Safety benchmarks such as HarmBench rely on LLM judges to classify model responses as harmful or safe, yet the judge configuration, namely the combination of judge model and judge prompt, is typically treated as a fixed implementation detail. We show this assumption is problematic. Using a 2 x 2 x 3 factorial design, we construct 12 judge prompt variants along two axes, evaluation structure and instruction framing, and apply them using a single judge model, Claude Sonnet 4-6, producing 28,812 judgments over six target models and 400 HarmBench behaviors. We find that prompt wording alone, holding the judge model fixed, shifts measured harmful-response rates by up to 24.2 percentage points, with even within-condition surface rewording causing swings of up to 20.1 percentage points. Model safety rankings are moderately unstable, with mean Kendall tau = 0.89, and category-level sensitivity ranges from 39.6 percentage points for copyright to 0 percentage points for harassment. A supplementary multi-judge experiment using three judge models shows that judge-model choice adds further variance. Our results demonstrate that judge prompt wording is a substantial, previously under-examined source of measurement variance in safety benchmarking.

cs.CL↗

GA-GS: Generation-Assisted Gaussian Splatting for Static Scene Reconstruction

Reconstructing static 3D scene from monocular video with dynamic objects is important for numerous applications such as virtual reality and autonomous driving. Current approaches typically rely on background for static scene reconstruction, limiting the ability to recover regions occluded by dynamic objects. In this paper, we propose GA-GS, a Generation-Assisted Gaussian Splatting method for Static Scene Reconstruction. The key innovation of our work lies in leveraging generation to assist in reconstructing occluded regions. We employ a motion-aware module to segment and remove dynamic regions, and thenuse a diffusion model to inpaint the occluded areas, providing pseudo-ground-truth supervision. To balance contributions from real background and generated region, we introduce a learnable authenticity scalar for each Gaussian primitive, which dynamically modulates opacity during splatting for authenticity-aware rendering and supervision. Since no existing dataset provides ground-truth static scene of video with dynamic objects, we construct a dataset named Trajectory-Match, using a fixed-path robot to record each scene with/without dynamic objects, enabling quantitative evaluation in reconstruction of occluded regions. Extensive experiments on both the DAVIS and our dataset show that GA-GS achieves state-of-the-art performance in static scene reconstruction, especially in challenging scenarios with large-scale, persistent occlusions.

cs.CV↗

Beyond Creed: A Non-Identity Safety Condition A Strong Empirical Alternative to Identity Framing in Low-Data LoRA Fine-Tuning

How safety supervision is written may matter more than the explicit identity content it contains. We study low-data LoRA safety fine-tuning with four supervision formats built from the same core safety rules: constitutional rules (A), creed-style identity framing (B), a B-matched creed condition with a worldview/confession identity-maintenance tail (C), and a matched non-identity condition (D). Across three instruction-tuned model families (Llama 3.1 8B, Qwen2.5 7B, and Gemma 3 4B), we evaluate HarmBench using a reconciled dual-judge pipeline combining Bedrock-hosted DeepSeek v3.2 and Sonnet 4.6, with disagreement and boundary cases manually resolved. The non-identity condition D is the strongest group on all three model families on the full 320-behavior HarmBench set, reaching 74.4% refusal on Llama, 76.9% on Gemma, and 74.1% on Qwen. By comparison, creed-style framing (B) improves over plain constitutional rules (A) on Llama and Gemma, but remains substantially below D, yielding an overall descriptive ordering of $D > B > C \geq A > baseline$. This provides a bounded empirical challenge to a strong version of the identity-framing hypothesis: explicit creed-style identity language is not necessary for the strongest gains observed here. Capability evaluations on MMLU and ARC-Challenge show no meaningful trade-off across conditions.

cs.CL↗

GameVerse: Can Vision-Language Models Learn from Video-based Reflection?

Human gameplay is a visually grounded interaction loop in which players act, reflect on failures, and watch tutorials to refine strategies. Can Vision-Language Models (VLMs) also learn from video-based reflection? We present GameVerse, a comprehensive video game benchmark that enables a reflective visual interaction loop. Moving beyond traditional fire-and-forget evaluations, it uses a novel reflect-and-retry paradigm to assess how VLMs internalize visual experience and improve policies. To facilitate systematic and scalable evaluation, we also introduce a cognitive hierarchical taxonomy spanning 15 globally popular games, dual action space for both semantic and GUI control, and milestone evaluation using advanced VLMs to quantify progress. Our experiments show that VLMs benefit from video-based reflection in varied settings, and perform best by combining failure trajectories and expert tutorials-a training-free analogue to reinforcement learning (RL) plus supervised fine-tuning (SFT).Our project page is available at https://gameverse-bench.github.io/ . Our code is available at https://github.com/THUSI-Lab/GameVerse .

cs.CV↗

MMSF: Multitask and Multimodal Supervised Framework for WSI Classification and Survival Analysis

Multimodal evidence is critical in computational pathology: gigapixel whole slide images capture tumor morphology, while patient-level clinical descriptors preserve complementary context for prognosis. Integrating such heterogeneous signals remains challenging because feature spaces exhibit distinct statistics and scales. We introduce MMSF, a multitask and multimodal supervised framework built on a linear-complexity MIL backbone that explicitly decomposes and fuses cross-modal information. MMSF comprises a graph feature extraction module embedding tissue topology at the patch level, a clinical data embedding module standardizing patient attributes, a feature fusion module aligning modality-shared and modality-specific representations, and a Mamba-based MIL encoder with multitask prediction heads. Experiments on CAMELYON16 and TCGA-NSCLC demonstrate 2.1--6.6\% accuracy and 2.2--6.9\% AUC improvements over competitive baselines, while evaluations on five TCGA survival cohorts yield 7.1--9.8\% C-index improvements compared with unimodal methods and 5.6--7.1\% over multimodal alternatives.

cs.CV↗

OpenNovelty: An LLM-powered Agentic System for Verifiable Scholarly Novelty Assessment

Evaluating novelty is critical yet challenging in peer review, as reviewers must assess submissions against a vast, rapidly evolving literature. This report presents OpenNovelty, an LLM-powered agentic system for transparent, evidence-based novelty analysis. The system operates through four phases: (1) extracting the core task and contribution claims to generate retrieval queries; (2) retrieving relevant prior work based on extracted queries via semantic search engine; (3) constructing a hierarchical taxonomy of core-task-related work and performing contribution-level full-text comparisons against each contribution; and (4) synthesizing all analyses into a structured novelty report with explicit citations and evidence snippets. Unlike naive LLM-based approaches, \textsc{OpenNovelty} grounds all assessments in retrieved real papers, ensuring verifiable judgments. We deploy our system on 500+ ICLR 2026 submissions with all reports publicly available on our website, and preliminary analysis suggests it can identify relevant prior work, including closely related papers that authors may overlook. OpenNovelty aims to empower the research community with a scalable tool that promotes fair, consistent, and evidence-backed peer review.

cs.IR↗

SpatialSplat: Efficient Semantic 3D from Sparse Unposed Images

A major breakthrough in 3D reconstruction is the feedforward paradigm to generate pixel-wise 3D points or Gaussian primitives from sparse, unposed images. To further incorporate semantics while avoiding the significant memory and storage costs of high-dimensional semantic features, existing methods extend this paradigm by associating each primitive with a compressed semantic feature vector. However, these methods have two major limitations: (a) the naively compressed feature compromises expressiveness, affecting the model's ability to capture fine-grained semantics, and (b) the pixel-wise primitive prediction introduces redundancy in overlapping areas, causing unnecessary memory overhead. To this end, we introduce \textbf{SpatialSplat}, a feedforward framework that produces redundancy-aware Gaussians and capitalizes on a dual-field semantic representation. Particularly, with the insight that primitives within the same instance exhibit high semantic consistency, we decompose the semantic representation into a coarse feature field that encodes uncompressed semantics with minimal primitives, and a fine-grained yet low-dimensional feature field that captures detailed inter-instance relationships. Moreover, we propose a selective Gaussian mechanism, which retains only essential Gaussians in the scene, effectively eliminating redundant primitives. Our proposed Spatialsplat learns accurate semantic information and detailed instances prior with more compact 3D Gaussians, making semantic 3D reconstruction more applicable. We conduct extensive experiments to evaluate our method, demonstrating a remarkable 60\% reduction in scene representation parameters while achieving superior performance over state-of-the-art methods. The code is available at https://github.com/shengyuuu/SpatialSplat.git

cs.CV↗