VeloGB10 – A GB10-specific inference engine

(github.com)

1 points | by colingauvin 10 hours ago ago

1 comments

  • colingauvin 10 hours ago ago

    Not my project but wanted to share. My 2x Spark config with Qwen 3.8 Flash went from 37 tps to ~90 tps (prose). It's running phenomenally well and really impressive.