Fork at Doubt
Following the answers a language model nearly gave, and deciding which branches are worth keeping.
Notes on AI infrastructure, product engineering, and whatever I’m building this week.
Following the answers a language model nearly gave, and deciding which branches are worth keeping.
EvryStay's backend tests, and the things we tried that didn't work.
Smarter & deeper MTP, MTPLX, and less CPU/GPU synchronisation on Qwen.
My paper "Less is More" has been accepted for presentation at the BID Workshop at ICPP 2026.
Monash DeepNeuron placed equal third in the 8th APAC HPC-AI Competition, competing against 49 university teams.
CUDA, modules, PBS Pro, storage, and build notes for NSCC Singapore's H100 cluster.
Scratch storage, Slurm, caches, and module notes for Monash University's M3 cluster.
Installing, serving, and benchmarking DeepSeek-R1 with SGLang across competition clusters.
The flags, couplings, failed NCCL experiments, and parallel layouts behind the final result.
Three months at Monash DeepNeuron trying to get a neural cellular automaton to learn gastrulation from a real light-sheet recording.
A 28×28 canvas that runs your MNIST model on whatever you draw, as you draw it. Now on TestPyPI as mnist-interactive.