teach sand to think, give sand a telescope π
PhD @cambridge_uni; building agentic research systems for physics & astronomy. prev @guardian data science.
I built a scale 3D map of the universe you can fly through, from Earth to the edge of the observable universe, including 192 million stars from Gaia, 15 million galaxies from DESI, 7,000 exoplanets, and more, stitched together.
Runs in your browser with @threejs!
https://t.co/a6TAtLiePJ
Could be to do with overthinking leading to in distribution behaviour, risk taking requires either acting with insufficient planning (ignkrance to the risk) or acting despite knowledge of the full risks, and being more risk adverse than risk accepting one would rarely try anything cool
Since yesterday: AlexNet training is at 4x up from 3x on Vulkan-Honeykrips on @omarchymac compared to CPU. Full pytorch operation coverage with 226/641 operations on PyTorch OpInfo GPU accelerated.
Testing 3d scanning with this ~$50 lidar sensor (ST VL53L9CX) attached to an iPhone.
Works pretty well for quick indoor scans using just ARKit poses.
PCB is open source from RemiFM: https://t.co/Ts8t88zGCY
Running pytorch on @OmarchyMac? Join my new project https://t.co/WipoNCamoY, a PyTorch backend for the Honeykrisp Vulkan GPU driver to accelerate deep learning on Apple Silicon Linux.
@ylecun@gabriberton 5 makes me wonder if the closed labs have tried exhaustively to scale JEPA and failed or whether they have a blind spot to it
Or don't have the expertise idk