Junchi Yu

Coupled Harness--Model Evolution for Scientific Equation Discovery

Pengze Li, Jiaquan Zhang, Encheng Su +8

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…

Decoupling Token Roles in Autoregressive Pretraining

Suqin Yuan, Runqi Lin, Kevin Qinghong Lin +4

Autoregressive pretraining increasingly draws on heterogeneous data, making it important to understand how a model learns from an individual token. The next-token prediction object…

RAP: Research Attention Prediction Reveals Target-Conditioned Evidence Acquisition Biases

Yingqian Wu, Jingcong Liang, Siyuan Wang +4

Large language models (LLMs) increasingly act as research agents, yet their ability to track shifts in research attention is difficult to evaluate because reviews and research idea…

CausalGame: Benchmarking Causal Thinking of LLM Agents in Games

Zhenhao Chen, Yongqiang Chen, Chenxi Liu +7

Building AI Scientist agents with Large Language Models (LLMs) has recently attracted growing attention. Since scientific discovery fundamentally relies on uncovering causal relati…

Measuring Epistemic Resilience of LLMs Under Misleading Medical Context

Hongjian Zhou, Xinyu Zou, Jinge Wu +19

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 increa…

A Language Model from 1913: Pretraining on Historical Text

Xiaoxi Luo, Zachary Shinnick, Niclas Griesshaber +5

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 produ…

cs.CV2026

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

Fabian Degen, Oishi Deb, Jindong Gu +4

Planning records define restrictions over geographic areas, but their source documents often provide only indirect spatial evidence rather than machine-readable boundaries. We intr…

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

Aoxi Liu, Yupeng Chen, James Oldfield +5

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…

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

Bohang Sun, Max Zhu, Francesco Caso +5

Diffusion large language models promise faster generation by refining many token positions in parallel, but this parallelism introduces a hidden control problem: which proposed tok…

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

Puning Yang, Junchi Yu, Qizhou Wang +3

Mitigating sensitive and harmful outputs is fundamental to ensuring safe deployment of LLMs. Existing approaches typically follow two paradigms: Knowledge Deletion (KD), which eras…

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

Sean Wu, Pan Lu, Yupeng Chen +7

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…

Safer by Diffusion, Broken by Context: Diffusion LLM's Safety Blessing and Its Failure Mode

Zeyuan He, Yupeng Chen, Lang Lin +7

Diffusion large language models (D-LLMs) offer an alternative to autoregressive LLMs (AR-LLMs) and have demonstrated advantages in generation efficiency. Beyond the utility benefit…

TDGNet: Hallucination Detection in Diffusion Language Models via Temporal Dynamic Graphs

Arshia Hemmat, Philip Torr, Yongqiang Chen +1

Diffusion language models (D-LLMs) offer parallel denoising and bidirectional context, but hallucination detection for D-LLMs remains underexplored. Prior detectors developed for a…