/journal
Mostly about the unglamorous half of building with language models: deciding whether the output actually got better, and what to do when it did not.
Everyone can generate. Far fewer can tell you whether the output got better. Notes on building a small evaluation set by hand, calibrating a scorer against it, and what the disagreements taught me.
Read the post →A personal reflection on growing up in India, living and working in Japan, and learning to navigate two cultures that shaped who I am.
Read the post →An overview of autonomous driving concepts, SAE J3016 automation levels, and core system components used in modern self-driving vehicles.
Read the post →Step-by-step guide to installing NVIDIA Deep Learning Dataset Synthesizer (NDDS) with Unreal Engine 4.22 for generating synthetic datasets for deep learning.
Read the post →Step-by-step guide to installing Unreal Engine 4 on Ubuntu, including Epic Games account setup, GitHub access, build steps, and useful shell helpers.
Read the post →A practical guide to using HackMD for writing Markdown notes, publishing them, collaborating live, and syncing with GitHub.
Read the post →