Ion Stoica

23

Papers

565

Total Citations

Papers (23)

From Crowdsourced Data to High-quality Benchmarks: Arena-Hard and Benchbuilder Pipeline

OR-Bench: An Over-Refusal Benchmark for Large Language Models

LLM-Assisted Code Cleaning For Training Accurate Code Generators

WorldModelBench: Judging Video Generation Models As World Models

GameArena: Evaluating LLM Reasoning through Live Computer Games

Establishing Best Practices in Building Rigorous Agentic Benchmarks

VisionArena: 230k Real World User-VLM Conversations with Preference Labels

HashAttention: Semantic Sparsity for Faster Inference

Twilight: Adaptive Attention Sparsity with Hierarchical Top-$p$ Pruning

Online Speculative Decoding

Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference

MuxServe: Flexible Spatial-Temporal Multiplexing for Multiple LLM Serving

Trustless Audits without Revealing Data or Models

Break the Sequential Dependency of LLM Inference Using Lookahead Decoding

R2E: Turning any Github Repository into a Programming Agent Environment

Context-Aware Streaming Perception in Dynamic Environments

Communication-efficient Distributed SGD with Sketching

RLlib Flow: Distributed Reinforcement Learning is a Dataflow Problem

Representing Long-Range Context for Graph Neural Networks with Global Attention

Accelerating Quadratic Optimization with Reinforcement Learning

Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena

RLlib: Abstractions for Distributed Reinforcement Learning

Population Based Augmentation: Efficient Learning of Augmentation Policy Schedules