Sitemap
A list of all the posts and pages found on the site. For you robots out there, there is an XML version available for digesting as well.
Pages
Posts
In the AI Era, Iteration Speed Is the Edge
Published:
In the AI era, how fast you iterate and how many projects you can move forward have become the skills that matter most. Three habits that speed me up: work like a manager, stay curious about new features, and match each task to the right model.
AI 时代，迭代速度就是竞争力
Published:
在 AI 时代，项目的迭代速度和数量正在成为最重要的能力。分享三条我自己的提速方法：多线程当管理者、保持对新功能的好奇、按任务选模型。
Deep Dive into speedup_by_fusion in PyTorch Inductor
Published:
A benchmark-driven analysis of PyTorch Inductor’s speedup_by_fusion config, its runtime logs, and why register spilling can reject fusions that still help.
PyTorch Inductor 中 speedup_by_fusion 深度解析
Published:
详解 PyTorch Inductor 的 speedup_by_fusion 配置：开启方式、工作原理、benchmark 日志示例，以及 register spilling 带来的融合决策争议。
Deep Dive into fuse_nodes in PyTorch Inductor
Published:
A structured walkthrough of how PyTorch Inductor enumerates, scores, and greedily applies graph fusion candidates in fuse_nodes.
PyTorch Inductor 中 fuse_nodes 融合流程深度解析
Published:
系统梳理 PyTorch Inductor 如何在 fuse_nodes 中枚举候选对、执行打分排序，并以贪心策略推进图融合。
portfolio
Portfolio item number 1
Short description of portfolio item number 1
Portfolio item number 2
Short description of portfolio item number 2
publications
TARE: Timescale-Aware Row-wise Energy Redistribution for Muon Optimizers
Published in Under review, 2026
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.
GraphFuse-R1: A Compiler-Side Bound and a Learned Graph-Level Policy for Operator Fusion
Published in Under review, 2026
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.
Syntactic Prediction through Reinforcement Learning
Published in Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026), Main Conference, 2026
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.
HR-Merge: Model Merging with Headroom-Weighted Residual Refinement
Published in NeurIPS 2026 Workshop on Neural Artifacts (NeuralArtifacts), Poster, 2026
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.
talks
Talk 1 on Relevant Topic in Your Field
Published:
This is a description of your talk, which is a markdown file that can be all markdown-ified like any other post. Yay markdown!
Conference Proceeding talk 3 on Relevant Topic in Your Field
Published:
This is a description of your conference proceedings talk, note the different field in type. You can put anything in this field.
teaching
Teaching experience 1
Undergraduate course, University 1, Department, 2014
This is a description of a teaching experience. You can use markdown like any other post.
Teaching experience 2
Workshop, University 1, Department, 2015
This is a description of a teaching experience. You can use markdown like any other post.