kunal51107/Speculative-decoding-engine

LLM inference engine built from scratch in PyTorch. Achieves 2-3x speedup via speculative sampling.

4 best points · 1 comments · 1 Hacker News post · 2 GitHub stars · View on GitHub

Hacker News discussions

• Show HN: Speculative Decoding from Scratch in PyTorch (2.8x CPU Speedup) (4 points, 1 comments, by kunal51107, 2025-11-30)