Show HN: Narwhal – LLM serving that moves GPUs between prefill/decode in seconds

3 comments

• z0nda 5 hours ago ago

  What happens to the model weights when a role flip takes place?

  [-]
  • athrael-soju 5 hours ago ago

    [flagged]

• athrael-soju 6 hours ago ago

  [flagged]