Laguna XS 2.1 33B on a RTX 3090: 296 tok/s peak, 152 tok/s at 256K contextlucebox.com 20GreenGames2mo1 comment
PFlash: 10x prefill speedup over llama.cpp at 128K on a RTX 3090github.com/Luce-Org 3GreenGames4mo1 comment
Show HN: OS Megakernel that match M5 Max Tok/w at 2x the Throughput on RTX 3090github.com/Luce-Org 6GreenGames5mo1 comment
Improvements in reasoning AI models may slow down soon, analysis findstechcrunch.com 3GreenGames1ydiscuss
Show HN: Tlume – a CLI tool that converts Tart VM images for Lumegithub.com/aktech 5GreenGames1y2 comments
Show HN: Lume – OS lightweight CLI for MacOS and Linux VMs on Apple Silicongithub.com/trycua 309GreenGames1y75 comments