Jiakang Huang

HR-Merge: Model Merging with Headroom-Weighted Residual Refinement

Jiakang Huang, Xueyan Zhang, Renyi Cai, Jingcheng Niu, Jiaru Li, Yining Wang, Josh Zhang, Jiahe Jiang, Jinman Zhao, Gerald Penn, Baharan Mirzasoleiman

NeurIPS 2026 NeuralArtifacts Workshop (Poster). A training-free, merge-operator-agnostic wrapper that provably never underperforms the base merge on validation. It gives the best result in every CLIP ViT and LoRA setting tested, reaching 94.29 on the LLaMA-3 8B six-NLI benchmark.

Syntactic Prediction through Reinforcement Learning

Jinman Zhao, Yining Wang, Jiahe Liu, Xueyan Zhang, Linbo Cao, Jiakang Huang, Yiren Zhao, Renyi Cai, YiTian Ding, Gerald Penn

Accepted to EMNLP 2026 (Main Conference). Syntax-R1 is a two-stage post-training framework that uses reinforcement learning to improve hierarchical syntactic prediction in large language models, achieving state-of-the-art results on constituency parsing and CCG supertagging.

GraphFuse-R1: A Compiler-Side Bound and a Learned Graph-Level Policy for Operator Fusion

Xueyan Zhang, Jiakang Huang, YiTian Ding, Shuhao Guan, Yining Wang, Yuchen Li, Jinman Zhao, Gerald Penn

Under review. GraphFuse-R1 trains an 8B LLM with SFT and GRPO to propose whole-graph operator-fusion groups inside PyTorch's TorchInductor, compiling 28.5% fewer kernels and fusing 39% more nodes than the hand-written heuristic, and derives a compiler-side bound on how much fusion headroom remains.

TARE: Timescale-Aware Row-wise Energy Redistribution for Muon Optimizers

Jinman Zhao, Xueyan Zhang, Jiakang Huang, Xutao Mao, Jingcheng Niu, Xi Chen, Yining Wang, Changyi Yang, YiTian Ding, Jiahe Liu, Gerald Penn, Jiaru Li

Under review. A Muon-family optimizer that shifts update energy away from weight rows whose recent gradients conflict with the longer-term trend, reaching NorMuon's loss with 1.07–1.10x higher token efficiency from 130M to 1.1B parameters at under 5% extra wall-clock time.

OpenTikZ

An open-source TikZ figure library and AI agent skill for academic papers. Describe a figure, such as a system diagram, neural-network architecture, or pipeline, and get back paper-ready, editable TikZ.

iMark

Co-founder & AI Technical Lead. An agentic bookmark assistant with semantic search and web interactions, shipped as a Chrome extension with 100+ users. Built on retrieval-augmented generation (Pinecone + Gemini), with Stripe and Supabase.

UBC Campus Explorer

A full-stack UBC campus navigation app: a custom query engine over UBC building and room data, with Google Maps walking routes between selected buildings (100+ weekly visits). Built with TypeScript, Express/Node.js, the Google Maps API, and Docker on AWS EC2.