15 papers
Asking What a Director Sees Can Change What Vision-Language Models Choose
Jiaxuan Guo, Xingyu Ren, Kejia Zhang +3
Vision-language models are now routinely tested on theory of mind: whether they can work out what another person sees, knows, and means. A common design asks the model to report th…
Scivot: Benchmarking Principal-Investigator Judgment at Pivotal Moments of Scientific AutoResearch Agents
Youran Sun, Xingyu Ren, Chugang Yi +15
Large language model (LLM) agents can keep scientific experiments running for months with little human supervision, but reliable autonomous research also requires scientific judgme…
When an Evaluation Rule Writes Training Labels: Measuring Human-Reference Forgiveness in NAVSIM
Jiaxuan Guo, Jingxin Yang, Jiaqi Ye +4
When the human reference scores zero on a metric, the released GTRS-Dense label generator for NAVSIM marks every candidate trajectory in the scene as passing it. NAVSIM's authors i…
Scoring the Wrong Question: Readout Failures in Constrained-Option Evaluation
Jiaxuan Guo, Kejia Zhang, Shuo Xin +3
Constrained-option scoring reads a model's probabilities for a fixed set of permitted answers, so it returns a score even when the model is about to write something else. We study…
AutoSR: Automatic Symbolic Regression by Searching Research States
Kejia Zhang, Youran Sun, Xinyu Ren +2
We introduce Automatic Symbolic Regression (AutoSR), a fully automated system that instantiates Research-Space Symbolic Regression by searching persistent scientific investigations…
Reward-Oracle MCTS for Formal Theorem Proving: Sample-Efficient Search and the Need for Kernel-Level Proof Auditing
Bodla Krishna Vamshi, Haizhao Yang
Formal theorem proving with large language models remains challenging due to the difficulty of navigating large proof search spaces efficiently. Existing tree search approaches eit…
AgonAlpha: Autonomous Alpha Discovery via Prompt Economy and Scalable Agentic Search
Weicheng Ye, Youran Sun, Xingyu Ren +3
Language models can propose many plausible trading factors, but an autonomous research system must also allocate its evaluation budget, verify its own evidence, and preserve how ea…
Understanding Sparse Attention Selectivity in Long-Context Foundation Models via Counterfactual Evaluation
Xingyu Ren, Youran Sun, Chugang Yi +1
Sparse attention is widely deployed in long-context serving stacks, yet no framework audits how discarding blocks changes the influence of specific content on model output. We firs…
Cross-Task Dissociation in Frontier Vision-Language Model Theory of Mind
Kejia Zhang, Youran Sun, Chugang Yi +1
Do frontier vision-language models present a coherent Theory-of-Mind (ToM) profile across tasks, matching the same human reference group, or does that profile fragment from one par…
Effective Parameters, Real Behavior: Renormalization for Robotics -- From Infinite Electron Mass to Sim-to-Real Gap
Youran Sun, Jiaxuan Guo, Xingyu Ren +2
Bridging the sim-to-real gap is a central problem in robotics, and the prevailing approach is to build increasingly accurate simulators. Here, we propose another approach based on…
Bringing Agentic Search to Earth Observation Data Discovery
Minghan Yu, Youran Sun, Chugang Yi +2
NASA and its data centers hold thousands of geoscience datasets and tools like Worldview, Giovanni, the Science Discovery Engine, and Harmony. Finding the right one is hard even fo…
Agon: An Autonomous Large-Scale Omnidisciplinary Research System Built on Prompt Economy
Youran Sun, Xingyu Ren, Chugang Yi +4
Large language models are making research production scalable, shifting the bottleneck from producing artifacts to judging claims. We present \textsc{Agon}, a research orchestrator…
ReSearch: A Multi-Stage Machine Learning Framework for Earth Science Data Discovery
Youran Sun, Yixin Wen, Haizhao Yang
The rapid expansion of Earth Science data from satellite observations, reanalysis products, and numerical simulations has created a critical bottleneck in scientific discovery, nam…
DRIFT: Detecting Representational Inconsistencies for Factual Truthfulness
Rohan Bhatnagar, Youran Sun, Chi Andrew Zhang +2
LLMs often produce fluent but incorrect answers, yet detecting such hallucinations typically requires multiple sampling passes or post-hoc verification, adding significant latency…
Geometry-Aware Hallucination Detection in Large Language Models
Bodla Krishna Vamshi, Rohan Bhatnagar, Haizhao Yang
Large language models (LLMs) frequently generate factually incorrect or unsupported content, commonly referred to as hallucinations. Prior work has explored decoding strategies, re…