A arxiv.org Lossless Acceleration of LLM via Adaptive N-Gram Parallel Decoding 136 points PaulHoule 2 years ago 23 comments
