James Zou

In-the-Flow Agentic System Optimization for Effective Planning and Tool Use

InfoTok: Adaptive Discrete Video Tokenizer via Information-Theoretic Compression

Can Language Models Discover Scaling Laws?

Textual Equilibrium Propagation for Deep Compound AI Systems

Can Small Training Runs Reliably Guide Data Curation? Rethinking Proxy-Model Practice

Inference-time scaling of diffusion models through classical search

Cost-of-Pass: An Economic Framework for Evaluating Language Models

TaTToo: Tool-Grounded Thinking PRM for Test-Time Scaling in Tabular Reasoning

Optimas: Optimizing Compound AI Systems with Globally Aligned Local Rewards

When Does Divide and Conquer Work for Long Context LLM? A Noise Decomposition Framework

reAR: Rethinking Visual Autoregressive Models via Token-wise Consistency Regularization

Cartridges: Lightweight and general-purpose long context representations via self-study

PRISM-Physics: Causal DAG-Based Process Evaluation for Physics Reasoning

Learning Human-Perceived Fakeness in AI-Generated Videos via Multimodal LLMs

ACT: AGENTIC CLASSIFICATION TREE

LLMSELECTOR: Towards Model Selection Optimization for Compound AI Systems

FACET: A Fragment-Aware Conformer Ensemble Transformer

Graph-of-Agents: A Graph-based Framework for Multi-Agent LLM Collaboration

Fractional Reasoning via Latent Steering Vectors Improves Inference Time Compute

How Well Can General Vision-Language Models Learn Medicine By Watching Public Educational Videos?

Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models

Impatient Users Confuse AI Agents: High-fidelity Simulations of Human Traits for Testing Agents

UQ: Assessing Language Models on Unsolved Questions

When Visualizing is the First Step to Reasoning: MIRA, a Benchmark for Visual Chain-of-Thought

Where LLM Agents Fail And How They can Learn From Failures

S-Chain: Structured Visual Chain-of-Thought for Medicine

Moloch's Bargain: Emergent Misalignment When LLMs Compete for Audience Success

MedCaseReasoning: Evaluating and learning diagnostic reasoning from clinical case reports

Think Deep, Think Fast: Investigating Efficiency of Trained-verifier-free Inference-time-scaling Methods

A Snapshot of Influence: A Local Data Attribution Framework for Online Reinforcement Learning

Capturing the Temporal Dependence of Training Data Influence

EvoLM: In Search of Lost Language Model Training Dynamics

CollabLLM: From Passive Responders to Active Collaborators

Mixture-of-Agents Enhances Large Language Model Capabilities

Reducing Hallucinations in Large Vision-Language Models via Latent Space Steering

metaTextGrad: Automatically optimizing language model optimizers

GeoAda: Efficiently Finetune Geometric Diffusion Models with Equivariant Adapters

Cost-efficient Collaboration between On-device and Cloud Language Models

MMed-RAG: Versatile Multimodal RAG System for Medical Vision Language Models

More Thinking, Less Seeing? Assessing Amplified Hallucination in Multimodal Reasoning Models

AutoRedTeamer: Autonomous Red Teaming with Lifelong Attack Integration

GMValuator: Similarity-based Data Valuation for Generative Models

4KAgent: Agentic Any Image to 4K Super-Resolution

ExGra-Med: Extended Context Graph Alignment for Medical Vision-Language Models

Weak-for-Strong: Training Weak Meta-Agent to Harness Strong Executors

FactTest: Factuality Testing in Large Language Models with Finite-Sample and Distribution-Free Guarantees

Cats Confuse Reasoning LLM: Query Agnostic Adversarial Triggers for Reasoning Models

Improving Model Alignment Through Collective Intelligence of Open-Source Models

Locality Alignment Improves Vision-Language Models

MedTrinity-25M: A Large-scale Multimodal Dataset with Multigranular Annotations for Medicine

SiriuS: Self-improving Multi-agent Systems via Bootstrapped Reasoning

Optimizing Calibration by Gaining Aware of Prediction Correctness

Dragonfly: Multi-Resolution Zoom-In Encoding Enhances Vision-Language Models

Quantifying AI Psychology: A Psychometric Benchmark for Large Language Models

LoGra-Med: Long-Context Multi-Graph Alignment for Medical Visual-Language Models

Self-rationalization improves LLM as a fine-grained judge

Improving Model Alignment Through Collective Intelligence of Open-Source Models

FactTest: Factuality Testing in Large Language Models with Statistical Guarantees

SMiR: A Synthetic Data Pipeline To Improve Multi-Image Reasoning

AutoRedTeamer: An Autonomous Red Teaming Agent Against Language Models