Search arXiv⌕ Search

arXiv subjects

Junchi Yu

Publications and source records attributed to Junchi Yu.

At least 19 recordsLinked to original sources
arXiv2026

Measuring AI Agents as Teammates with Collaborative Intelligence

Progress in AI is often measured by what a single model can do on its own. Yet teams of AI agents are starting to make breakthroughs and discoveries in science, medicine, mathematics and software engineering. As AI begins to work in teams of agents and people, the ability to collaborate becomes a core part of their intelligence. Research on human teams has measured this ability for decades. Inspired by this, we introduce the Collaborative Intelligence Benchmark, which measures how well an AI agent works as a teammate. It contains 240 tasks across 8 dimensions of teamwork, from refusing an unsafe shortcut to checking the work that a weaker model leaves behind. Each task comes from a real record or an existing benchmark, from aviation accident reports and clinical cases to coding tasks and negotiations, and has a programmatically verifiable outcome. To isolate teamwork from individual capability, models are scored only on tasks that they can solve by themselves when they have full information. We evaluate 18 closed- and open-weight models and find that every model, across sizes and families, has clear weaknesses as a teammate, with the best model scoring only 82.2 out of 100. Frontier models almost always refuse unsafe shortcuts and report misconduct, yet they cave under interpersonal pressure, reaching the best overall deal in only 14% of conflicts, and their pass rate drops from 97% to 81% when they take over a task in which a weaker model has cheated. Crucially, general capability does not guarantee good teamwork, as models with nearly the same capability score differ by up to 11 points as teammates. The benchmark includes 160 public tasks with code and data, along with 80 held-out tasks for evaluation. As AI agents increasingly work alongside other agents and people, the ability to collaborate effectively is becoming a crucial capability to measure and develop.

cs.AI↗
arXiv2026

RAP: Research Attention Prediction in Language Agents

LLM agents increasingly support research planning and execution, where judgments about current and anticipated research activity inform decisions. Yet whether they can infer changes across a whole field from the partial evidence they retrieve through online scholarly search remains unclear. We introduce Research Attention Prediction (RAP), a literature search benchmark for evaluating this ability. We build a local search environment from 356,357 dated arXiv papers and construct two complementary evaluation tasks. The benchmark comprises a Core set of 2,160 questions across 54 AI/ML fields and an Extended set covering 269 fields. Our evaluation and analysis of LLM agents show that search substantially improves stronger agents' recognition of recent changes, while forecasting gains remain limited or negative. Agents update forecasts across dates and adjust their investigations to forecasting goals, but these adaptations do not reliably yield accurate predictions of future shifts. Even for recent changes, final judgments can diverge from what the retrieved papers indicate. These findings show that agents can recognise some recent changes, yet reliably anticipating future research activity remains difficult despite changes in their answers and investigations.

cs.AI↗
arXiv2026

UnlearningSoup: Is Repeated Tuning Necessary for Large Language Model Unlearning?

Large language models trained on vast corpora inherently risk memorizing harmful content that may later re-emerge in their outputs. To mitigate this issue, existing unlearning methods typically rely on training-based parameter updates, such as gradient ascent and its variants, to delete targeted content while preserving other knowledge. However, balancing the competing goals of forgetting and retention makes hyperparameter choices for these methods particularly difficult, often requiring repeated tuning to obtain a strong model that still leaves substantial room for improvement and transfers poorly across models and datasets. To address this challenge, we investigate whether unlearning runs exhibit exploitable structure in weight space, and observe that models from different runs still lie in a shared evaluation-performance basin. This suggests that stronger models may be recovered through an unlearning-tailored soup strategy, reducing the need for repeated tuning for further improvement or new settings. Motivated by this, we propose UnlearningSoup, a unified framework that provides two strategies: EfficientSoup uses binary-search-based interpolation to quickly discover a well-performing model in the early stage, where repeated tuning would otherwise make strong model selection costly. PerformanceSoup uses reweighted souping to efficiently unlock the remaining performance potential in the later stage, where repeated tuning becomes increasingly inefficient. Extensive experiments across diverse datasets and models show that UnlearningSoup delivers 2.4x to 3.3x efficiency gains in hyperparameter selection, while consistently improving performance across settings.

cs.LG↗
arXiv2026

Coupled Harness--Model Evolution for Scientific Equation Discovery

Large language models (LLMs) are increasingly used to discover scientific equations from experimental data. Such a discovery system has two parts that must work together: a library of executable computational tools and the model's tool-use policy. The tools supply the solving capability, from diagnosing structure in the data to fitting candidate equations; the policy plans and orchestrates their use. Yet existing systems adapt only one part, keeping either the toolset or the model fixed, which leaves the corresponding failures uncorrectable. We develop a coupled evolution framework that improves both parts from the same verifiable fit-error signal. In the tool-evolution phase, an agent grows an executable tool library from a small set of generic numerical operations by diagnosing its own failures: it authors a new tool when a capability is missing and revises its selection rules when the right tool went unused. In the policy-evolution phase, a small open model (roughly 3B active parameters) is fine-tuned on successful tool-use trajectories collected with the frozen library, on training formulas screened to be disjoint from all evaluation benchmarks. Ablations show the two phases are complementary: with no tool library at all the same model solves almost nothing, the tools account for most of the gain, and trajectory training adds a further gain that is largest on the most difficult problems. The full system solves 79.1/71.3\% of LSR-Synth problems at fit tolerances $10^{-2}$/$10^{-3}$, reaches 63.8\% pooled accuracy on the distributionally different SRSD-Feynman panel, and obtains a normalized CP3-28 aggregate of $-7.27$. These results suggest a broader recipe for scientific agents: let the model and its instruments improve each other under verifiable feedback.

cs.AI↗
arXiv2026

A Language Model from 1913: Pretraining on Historical Text

While modern language models increasingly rely on ever-larger web corpora, we show that pretraining on historical text (e.g., pre-1913 text) in a data-constrained setting can produce a temporally grounded language model that still shows reasonable performance on language understanding. However, developing History LMs requires addressing challenges in data quality, preventing temporal leakage in post-training, and constructing temporally aligned evaluations. We address these challenges and pretrain TypewriterLM, a 7.24B-parameter model with a 1913 knowledge cutoff. We construct TypewriterCorpus, a 54B-token historical corpus with extensive temporal filtering, propose lexically grounded instruction tuning that constrains all responses to vocabulary from historical source documents, and introduce History-Event, a benchmark of 2,344 events for evaluating both competence and cutoff adherence. We release TypewriterLM and all associated resources to support future research on History LMs.

cs.CL↗
arXiv2026

$D^2$-Monitor: Dynamic Safety Monitoring for Diffusion LLMs via Hesitation Signals

Despite the emergence of diffusion large language models (D-LLMs) as an alternative to autoregressive large language models (AR-LLMs), safety monitoring for D-LLMs remains largely unexplored. Unlike AR-LLMs, D-LLMs generate text through a multi-step denoising process, exposing intermediate hidden representations that may contain safety-relevant information unavailable in standard single-step monitoring setups. Motivated by the suitability of lightweight probes for always-on monitoring, we analyze which trajectory-level signals best indicate when such probes are likely to struggle. We find that the most informative signal is safety hesitation: intermediate hidden states repeatedly falling within a small margin of the probe's decision boundary. The number of such hesitation steps in D-LLM's trajectory predicts probe failure effectively, providing a proxy for sample difficulty. Building on this analysis, we propose $D^2$-Monitor, in which a lightweight probe runs on every sample, producing a base prediction and a hesitation estimate that decides whether to escalate the sample to a more expressive probe. That probe is trained on trajectories with at least one hesitation step, and reads only the minimal span of those steps. Evaluated on 3 datasets (WildGuardMix, ToxicChat, OpenAI-Moderation) across 4 D-LLMs, $D^2$-Monitor achieves state-of-the-art performance with a compact parameter footprint ($\leq$ 0.93M parameters), and exhibits the best trade-off between effectiveness and efficiency relative to 8 baselines.

cs.AI↗
arXiv2026

Decoupling Token Roles in Autoregressive Pretraining

Autoregressive pretraining increasingly draws on heterogeneous data, making it important to understand how a model learns from an individual token. The next-token prediction objective naturally identifies a token's contribution with its own loss. However, each token is not only a prediction target but also context for what follows. Using controlled corruption, we decouple these two roles and find a reversal: making a noisy token easier to predict reduces its damage as a target but increases it as context. The same decoupling helps explain text generated by language models: generation selects each token by its fit to the prefix, while its role as context is never tested against an independently determined continuation, because that continuation is generated to fit it. At known corrupted positions, acting through the context can reduce damage that removing the token's own loss does not. Understanding and controlling what a model learns from a token therefore requires decoupling its roles.

cs.LG↗
arXiv2026

PaperDoctor: Evidence-Grounded and Actionable Feedback for Scientific Papers in Progress

Autoresearch agents are reshaping the research ecosystem, but they can also let flawed claims enter the literature at scale. Human advisors catch such issues in drafts through careful, traceable feedback, yet advisor-style assessment requires extensive manual effort and does not scale. To shift automated paper assessment from a judge to a diagnostician, we introduce PaperDoctor, an agent framework for pre-submission feedback with three key innovations. First, a holistic hierarchical framework evaluates writing, layout, references, code, theory, prior work, and experiments through three layers: L1 surface screening, L2 typed verifiers that route each claim to the appropriate evidence, and L3 reproducers that rerun experiments by priority. Second, each finding contains an observation, a pointer to specific evidence such as a sentence, equation, or code line, and a revision suggestion, making critiques auditable and actionable. Third, PaperDoctor selectively rebuilds and reruns experiments based on claim importance and compute budget, surfacing reproducibility gaps and quantitative limitations that are invisible from the manuscript alone. We evaluate PaperDoctor on 30 in-progress papers, yielding 70.6% agreement and all positive holistic scores, and on 40 manuscripts across machine learning, natural science, and social science, covering human- and AI-authored papers with code. Overall, PaperDoctor produces more auditable feedback than human and other agentic reviewers, pairs critiques with concrete suggestions by design, and complements dimensions often overlooked by human reviewers. We also develop an interactive interface that lets authors browse findings grounded in their paper. PaperDoctor reframes automated paper assessment as diagnosis rather than verdict, taking a concrete step toward AI advisors for more rigorous AI-assisted scientific discovery.

cs.CL↗
arXiv2026

Scientific reasoning does not reliably translate into scientific forecasting in frontier AI

AI systems are increasingly used to support forward-looking scientific judgment, but it remains unclear whether they can form reliable expectations about future scientific advances. Here we show that strong scientific reasoning does not reliably translate into accurate forecasting of future scientific advances. To study this question, we introduce CUSP, a temporally grounded evaluation suite for event-level scientific forecasting across eight scientific disciplines. Across six frontier AI models, we observe a striking asymmetry in forecasting performance together with systematic error patterns. Models often identify plausible mechanisms underlying future scientific advances, yet perform near chance on feasibility assessment, generate solution strategies that only weakly align with realized advances, and systematically predict scientific advances later than they become publicly observable. Providing additional pre-cutoff scientific knowledge improves performance but does not eliminate these forecasting limitations. These findings suggest that current AI systems possess substantial retrospective scientific competence but limited forward-looking predictive capability. Scientific forecasting should therefore be evaluated as a complementary dimension of AI scientific capability when deploying AI systems for research prioritization and scientific decision-making.

cs.AI↗
arXiv2026

FARS: A Fully Automated Research System Deployed at Scale

Recent automated research systems show that language-model agents can generate hypotheses, run experiments, and write complete manuscripts, but most evidence still comes from selected examples, human-framed topics, or a few pre-defined research tasks. We present FARS (Fully Automated Research System), a fully automated AI-for-AI research system designed to operate across research topics at scale. FARS autonomously generates and advances projects through ideation, planning, experimentation, and writing, using stage-specific agents coordinated through a shared workspace that records proposals, code, logs, results, and manuscripts. In its first public deployment, FARS produced 166 complete research papers spanning 67 fine-grained AI/ML topics while preserving intermediate artifacts as an auditable corpus rather than a curated set of successes. We evaluate this corpus with 282 structured reviews from volunteer reviewers covering 140 papers, including overall ratings, sub-scores, integrity checks, and LLM-use disclosure. The reviews indicate that FARS can produce review-worthy and occasionally strong AI/ML research artifacts in a large-scale public deployment, while also exposing recurring failure modes in narrow experimental scope, methodological limitations, and integrity issues.

cs.AI↗
arXiv2026

CausalGame: Benchmarking Causal Thinking of LLM Agents in Games

Building AI Scientist agents with Large Language Models (LLMs) has recently attracted growing attention. Since scientific discovery fundamentally relies on uncovering causal relationships from observations, the capability of causal thinking, i.e., distinguishing causation from correlation and recognizing hidden biases, is essential to LLM agents. Although a number of benchmarks exist for AI Scientists, none explicitly incorporate challenges from selection bias, measurement error, and hidden confounders that widely exist in real-world scientific discovery. To this end, we present CausalGame, a benchmark that evaluates the causal thinking capabilities of LLM agents through interactive games. CausalGame asks LLM agents to actively design experimental protocols, collect observation data, and derive a final solution with an explanation report. To emulate realistic scientific discovery challenges, we design 14 scenarios that incorporate selection bias, measurement error, and hidden confounders. Across 30 LLM agents, none demonstrates reliable causal thinking: the best model reaches only 68.0% survival against analytical optima of 78-85%, and merely 5-7% of sessions receive credits on the causal-reasoning rubrics. CausalGame provides a scalable and controlled testbed for evaluating the causal thinking of AI Scientist agents.

cs.CL↗
arXiv2026

Measuring Epistemic Resilience of LLMs Under Misleading Medical Context

Large language models (LLMs) now reach expert-level scores on medical licensing exams, encouraging the assumption that high scores imply safe medical judgment while patients increasingly use them for health advice. We show this assumption is fragile: when misleading context is injected into questions that LLMs originally answer correctly, they abandon the correct answer. We call the ability to maintain correct judgment under adversarial context epistemic resilience, and introduce MedMisBench to measure it. MedMisBench contains 10,932 medical question items and 48,889 misleading context-option pairs spanning medical reasoning, agentic capability, and patient-journey evaluation. Across 11 model configurations, mean accuracy falls from 71.1% on original questions to 38.0% under focused misleading context, with 51.5% attack success. The most damaging injections are formal, rule-like fabrications: authority-framed falsehoods reach 69.5% attack success and exception-poisoning claims reach 64.1%. A 14-member clinical panel from 7 countries identified serious potential harm in 38.2% of reviewed cases. MedMisBench exposes a structural blind spot in LLM evaluation in medical settings: existing benchmarks measure what models know, but not whether they preserve correct medical judgment under misleading context.

cs.CL↗
arXiv2026

Plan2Map: A Multimodal Benchmark for Document-Grounded Geospatial Boundary Reconstruction from Planning Records

Planning records define restrictions over geographic areas, but their source documents often provide only indirect spatial evidence rather than machine-readable boundaries. We introduce Plan2Map, a 208-case multimodal benchmark for document-grounded geospatial boundary reconstruction from UK planning records. Given only a source planning document, systems must reconstruct a valid geospatial boundary from notice text, schedules, map plates, map labels, and boundary annotations; the reference GeoJSON is held out for scoring. We propose GeoPlanAgent, a document-grounded, geospatial-tool-in-the-loop system that decomposes the task into evidence extraction, localisation, map registration, boundary segmentation, projection, and verification. On Plan2Map, GeoPlanAgent achieves 0.736 mean IoU and 0.904 median IoU, with 67.8\% of predictions at or above 0.8 IoU, substantially outperforming direct VLM-to-GeoJSON baselines. Diagnostic analysis shows that direct VLM prediction remains unreliable, while remaining errors are concentrated in localisation and map registration, and supervised boundary segmentation substantially improves pixel-level mask quality. Plan2Map provides a concrete testbed for multimodal geospatial reconstruction from public planning records. Project page: https://odeb1.github.io/Plan2Map_Project_Page/.

cs.CV↗
arXiv2026

The Alignment Curse: Modality Alignment Supercharges Audio Attacks via Text Transfer

Recent advances in end-to-end trained omni-models have substantially improved audio capabilities by strengthening text-audio modality alignment. However, whether such alignment inadvertently facilitates the transfer of safety vulnerabilities across modalities remains underexplored. This question is critical as text-based jailbreak attacks are considerably more mature than audio-based ones; if they transfer systematically, current audio safety evaluations may underestimate risks originating from the text modality. In this paper, we introduce the Alignment Curse, a formally characterized and empirically validated principle showing that stronger modality alignment enables more effective transfer of attacks from text to audio, revealing a fundamental tension between capability and safety. Motivated by this principle, we conduct a comprehensive black-box evaluation of three attack categories on recent omni-models (e.g., Qwen2.5-Omni, Qwen3-Omni): text attacks, text-transferred audio attacks, and audio attacks. We find that text-transferred audio attacks perform comparably to, and often better than, audio-based attacks, exhibiting a clear advantage under audio-only access. This suggests that text-based vulnerabilities play a pivotal role in shaping audio safety risks. Finally, we empirically analyze the relationship between modality alignment and transfer effectiveness across attack methods and models, observing consistent support for the Alignment Curse: tighter modality alignment leads to more effective cross-modality attack transfer.

cs.LG↗
arXiv2026

The Path Matters: Learning a Token-Commitment Policy for Diffusion Language Models

Diffusion large language models promise faster generation by refining many token positions in parallel, but this parallelism introduces a hidden control problem: which proposed tokens should be transferred into the partially decoded sequence at each step? We refer to this decision as token commitment. Existing frozen-generator decoders largely rely on hand-designed confidence rules or block-specific acceptance filters. We argue that token commitment can instead be learned as a reusable trace-state policy. We introduce TraceLock, a lightweight plug-in controller that instantiates this policy for a frozen diffusion language model. Since oracle commitment times are unavailable, TraceLock derives self-supervision from future stability: at decoding step t, a proposed token for position i is labeled stable if it matches the final token at position i after the full decoding trace completes. The controller scores variable-length trace states and decides which active token proposals should be committed to the partially decoded sequence. Once trained for a given frozen backbone, the controller can be deployed across local-window widths, generation lengths, and step budgets without retraining or per-setting calibration. Experiments on question answering, mathematical reasoning, and code generation show that TraceLock improves the quality-step tradeoff over heuristic and learned baselines, with particularly stable behavior under cross-setting deployment. Diagnostic analyses show that its decisions are not reducible to scalar confidence, suggesting that frozen diffusion language models expose a learnable space of commitment trajectories beyond confidence-based decoding. Code is available at https://github.com/BobSun98/TraceLock.

cs.CL↗
arXiv2026

Distinguishable Deletion: Unifying Knowledge Erasure and Refusal for Large Language Model Unlearning

Mitigating sensitive and harmful outputs is fundamental to ensuring safe deployment of LLMs. Existing approaches typically follow two paradigms: Knowledge Deletion (KD), which erases undesirable information during training, and Distinguishable Refusal (DR), which steers models away from using sensitive knowledge during inference. Despite rapid progress, KD-based unlearning struggles with biased deletion due to suppressing specific token sequences as a substitute for complete knowledge removal, whereas DR-based unlearning risks the re-emergence of harmful knowledge because the underlying knowledge remains intact. To address these issues, we propose Distinguishable Deletion ($\mathrm{D^2}$), a paradigm that restricts the response distribution in the latent representation rather than specific tokens to erase undesirable knowledge, while distinguishing it from retained knowledge, enabling a refusal mechanism to handle unlearned inputs safely and coherently. To implement $\mathrm{D^2}$, we introduce an energy index that quantifies the presence of knowledge and the separation between unlearned and retained content. Mathematical and empirical analyses show that energy is both accurate and efficient, enabling Energy-based Unlearning Alignment (EUA) to enforce energy-boundary unlearning during training and apply an energy-based refusal mechanism at inference. Extensive experiments demonstrate that EUA significantly outperforms previous methods, indicating the superiority of $\mathrm{D^2}$. Our code is available at https://github.com/Puning97/EUA-for-LLM-Unlearning.

cs.LG↗
arXiv2026

Causal Fine-Tuning under Latent Confounded Shift

Adapting to latent confounded shift remains a core challenge in modern AI. This setting is driven by hidden variables that induce spurious correlations between inputs and outputs during training, leading models to rely on non-causal shortcuts. For example, a model may learn to treat metadata (e.g., data source like "Amazon") as a proxy for positive sentiment, causing failure when the source becomes predominantly negative during deployment. To address this latent confounded shift, we introduce Causal Fine-Tuning(CFT). Using a structural causal model as an inductive bias, we derive sufficient identification conditions that motivate a fine-tuning objective for decomposing representations into high-level stable and low-level shift-sensitive components. Instantiating this framework in BERT, we show that learning such causal/spurious representations and adjusting them accordingly yield a more robust predictor. Experiments on spurious correlation injection attacks in text demonstrate that our method outperforms black-box domain generalization baselines, highlighting the benefits of explicitly modeling causal structure.

cs.LG↗
arXiv2026

Hidden in Plain Sight: Visual-to-Symbolic Analytical Solution Inference from Field Visualizations

Recovering analytical solutions of physical fields from visual observations is a fundamental yet underexplored capability for AI-assisted scientific reasoning. We study visual-to-symbolic analytical solution inference (ViSA) for two-dimensional linear steady-state fields: given field visualizations (and first-order derivatives) plus minimal auxiliary metadata, the model must output a single executable SymPy expression with fully instantiated numeric constants. We introduce ViSA-R2 and align it with a self-verifying, solution-centric chain-of-thought pipeline that follows a physicist-like pathway: structural pattern recognition solution-family (ansatz) hypothesis parameter derivation consistency verification. We also release ViSA-Bench, a VLM-ready synthetic benchmark covering 30 linear steady-state scenarios with verifiable analytical/symbolic annotations, and evaluate predictions by numerical accuracy, expression-structure similarity, and character-level accuracy. Using an 8B open-weight Qwen3-VL backbone, ViSA-R2 outperforms strong open-source baselines and the evaluated closed-source frontier VLMs under a standardized protocol.

cs.AI↗