NihilDigit

I'm a third-year undergraduate at North China University of Science and Technology, studying intelligence science. I use frontier AI to push into the problems that genuinely excite me.

ANVIL grew out of two threads I'd been pulling on: I'm picky about frame rate, and I'd been tinkering with mobile NPUs. Standard flow-based VFI breaks down under INT8 quantization, so I went looking for alternatives and landed on reusing the H.264 decoder's motion vectors instead of learning optical flow. The paper is under review at IEEE TCSVT; the project page collects the training code, the Android demo, and the arXiv preprint.

RAFNet was my first academic project. We built a classifier for student behavior in crowded classrooms on top of GroupRec, an existing hypergraph encoder, with a gated fusion head; I ran the experiments. It's under review at The Visual Computer.

I want my code to do something for people, not just sit in a repo. A fair amount of free time goes to open source — most visibly waybar-ai-usage, a Waybar widget that shows quota and reset time for Claude Code, Codex CLI, Copilot, and OpenCode Zen. It's packaged on AUR.

A short list of things I'm into: LLMs, Linux, ACGN, Rammstein. Beyond just using them, I find different language models genuinely fascinating — each one has its own character, and I spend a lot of hours just talking to them. ACGN bleeds into what I build: an upstream PR in review for cloud offline-download in Animeko (an open-source anime streaming app), and renpak, an AVIF compression toolchain for Ren'Py visual novels.