webgpu
2 posts
Article
MicroLLM Lab runs seven models without touching the network
MicroLLM Lab runs seven 25M-360M parameter language models in the browser via WebGPU, on-device with no server, and benchmarks their speed and accuracy.
Article
A single file freezes your Mac
A WebGPU shader with an infinite loop locks up an M-series Mac until a watchdog forces a reboot. One click triggers it, and Apple declines to fix it.