笔记
研究笔记、工程经验与进行中的想法。
关于机器学习可靠性、数值调试、GPU 可移植性、软件工程,以及搭建科研系统时实际取舍的简短技术文章。
文章正文为英文。
2026年5月1日
What I Want This Blog to Become
Academic papers are necessarily compressed. They describe the method, the evaluation, and the result, but they often leave out the messy middle: why a test design changed, why o...
阅读笔记 → 2025年7月8日Automatically Detecting Numerical Instability in ML via Soft Assertions
In our FSE 2025 paper, we introduce Soft Assertions — a new approach to detect and trigger these hidden bugs automatically.
阅读笔记 → 2025年7月7日Testing GPU Numerics: Finding Numerical Differences Between NVIDIA and AMD GPUs
Our SC24-W workshop paper presents a systematic method to detect and analyze these cross-vendor numerical differences.
阅读笔记 →