Svelte Hacker News logo
  • top
  • new
  • show
  • ask
  • jobs
  • about

Eagle-3 Speculative Decoding for LLM Inference (5.6x speedup)

github.com

2 points by summarity 3 days ago