AirLLM
Runs huge models on small GPUs by streaming layers
Real-time 3D scene reconstruction from camera frames
An open-source model that reconstructs a 3D scene in real time from a stream of camera frames. It is for researchers and creators exploring computer vision, mapping, and 3D capture.
Takes streaming video frames and pieces together a 3D map, showing the input video alongside the map being built. The video notes it requires an Nvidia GPU, though resulting artifacts can run on more efficient hardware.
Free software: Google, Notion, Supabase, etc. GitHub’s top 10 repos
Runs huge models on small GPUs by streaming layers
Public IPTV channel playlists ready for VLC and other players
Real-time face swap from a single reference photo
Self-hosted music hub for smart-home audio and streaming services
Build a ChatGPT-like LLM in PyTorch step by step
Train a tiny language model from scratch on regular hardware