Zhuoming Chen

FSA: An Alternative Efficient Implementation of Native Sparse Attention Kernel

Jackpot: Align Actor-Policy Distribution for scalable and stable RL for LLM

WWW.Serve: A Decentralized Framework for Collaborative LLM Serving

Kinetics: Rethinking Test-Time Scaling Law

MagicPIG: LSH Sampling for Efficient LLM Generation

MagicDec: Breaking the Latency-Throughput Tradeoff for Long Context Generation with Speculative Decoding

GSM-$\infty$: How Do your LLMs Behave over Infinitely Increasing Reasoning Complexity and Context Length?

FACTOR: Factoring Complexity and Context Length in Long-Context Model Evaluation

CAT Pruning: Cluster-Aware Token Pruning For Text-to-Image Diffusion Models