LLM Digest

Story

arxiv_cs_lg · Nov 3, 2026 · paper

Source brief

Give the Drafter a Second Chance to Guess: Parallel Reconsideration for Speculative Decoding

arxiv.orgNov 3, 2026
original source linked

In brief

Speculative decoding accelerates large language model inference with a small drafter, whose per-guess acceptance rate determines the achievable speedup. Feature-level drafters share a design choice: every token, easy...

Read the original at arxiv.org →Open in live feed Read that day’s brief

Give the Drafter a Second Chance to Guess: Parallel Reconsideration for Speculative Decoding

Earlier in this thread 4 items

Training Parallel Speculative Draft Models by Directly Minimizing Expected Decoding Rounds

NgramRL: Accelerating RL Post-Training for Structured Generation via Rollout-Aware N-gram Speculative Decoding

Secure Speculative Decoding for Large Language Models

Flash-dLLM: IO-Aware KV Caching and Parallel Decoding for Fast, Memory-Efficient Diffusion LLMs