Jiaping Wang’s HomePage

I received my bachelor’s degree in Software Engineering from East China Normal University in 2023 and master degree in 2026. My supervisor is Professor Jianwen Li.

My research primarily focuses on large language model (LLM) optimization, encompassing both training and inference. I am particularly interested in accelerating LLM training and inference through techniques such as speculative decoding and framework-level improvements. Now I’m maintaining the SpecForge, with the goal of enabling everyone to train the most efficient draft models for the latest models. Welcome anyone with shared interests in these topics or LLM to reach out and connect.

From May 2024 to March 2025, I interned in the Foundation Model Group at SenseTime Research. I worked on the reasoning acceleration of large models and data synthesis of foundation models.

Since May 2025, I have been with the Alibaba Cloud Compiler Group, working on LLM inference framework optimization and AI compiler development. I‘m also open to exciting career opportunities now.

News______________________________________________________

• ACL2025 Findings accepted our paper “Consultant Decoding: Yet Another Synergistic Mechanism,”,2025
• New preprint on “LeetDecoding: A PyTorch Library for Exponentially Decaying Causal Linear Attention with CUDA Implementations”,2025[arxiv]
• New preprint on “CFP: A Reinforcement Learning Framework for Comprehensive Fairness-Performance Trade-Off in Machine Learning”,2024[arxiv]
• New preprint on “Experimenting a new programming practice with llms”,2024[arxiv]