Laguna-XS-2.1-DFlash

Papers using or citing this model 1

• DFlash: Block Diffusion for Flash Speculative Decoding 144 citations · 2026 · arXiv (Cornell University) · arXiv

  Autoregressive large language models (LLMs) deliver strong performance but require inherently sequential decoding, leading to high inference latency and poor GPU utilization. Speculative decoding mitigates this bottleneck by using a...