Ben Athiwaratkun

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

CARE: Covariance-Aware and Rank-Enhanced Decomposition for Enabling Multi-Head Latent Attention

Support Your Local LMs: Redistributing LM Traffic from Cloud to Edge with TrafficBench

Understanding and Steering the Cognitive Behaviors of Reasoning Models at Test-Time

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

Imitate Optimal Policy: Prevail and Induce Action Collapse in Policy Gradient

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

Mixture-of-Agents Enhances Large Language Model Capabilities

Training-Free Activation Sparsity in Large Language Models

Weaver: Shrinking the Generation-Verification Gap by Scaling Compute for Verification

Improving Model Alignment Through Collective Intelligence of Open-Source Models

Scaling Instruction-tuned LLMs to Million-token Contexts via Hierarchical Synthetic Data Generation

Ladder-Residual: Parallelism-Aware Architecture for Accelerating Large Model Inference with Communication Overlapping

The Promises and Pitfalls of Language Models for Structured Numerical Data

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

Ladder Residual: Redefining Tensor Parallelism in Transformers for Accelerated Inference

Improving Model Alignment Through Collective Intelligence of Open-Source Models