Zihan Liu

4 papers MBZUAI IFM

• AceReason-Nemotron: Advancing Math and Code Reasoning through Reinforcement Learning
• Activation Control for Efficiently Eliciting Long Chain-of-thought Ability of Language Models
• ClusterFusion: Expanding Operator Fusion Scope for LLM Inference via Cluster-Level Collective Primitive
• Yggdrasil: Bridging Dynamic Speculation and Static Runtime for Latency-Optimal Tree-Based LLM Decoding