Show HN: Narwhal – LLM serving that moves GPUs between prefill/decode in seconds (github.com) 1 points by athrael-soju 19d ago ↗ HN
[–] z0nda 19d ago ↗ What happens to the model weights when a role flip takes place? [–] [dead] athrael-soju 19d ago ↗ [flagged]
3 comments
[ 6.0 ms ] story [ 12.8 ms ] thread