Researchers propose DeLS-Spec, a speculative decoding method that decouples long- and short-context experts to accelerate LLM inference. By adding a lightweight local head to the existing DFlash model…
DeepSeek-AI released DeepSpec, an open-source codebase for training and evaluating draft models for speculative decoding, supporting three draft model algorithms (DSpark, DFlash, Eagle3) and requiring…
DeepSeek open-sourced DSpark, a speculative decoding framework that accelerates per-user generation for DeepSeek-V4 by 57–85% over the MTP-1 baseline. The framework pairs a parallel draft backbone wit…
DeepSeek released DSpark, a speculative decoding framework now live in its DeepSeek-V4 Flash and Pro production API, delivering 51 to 400 percent throughput gains and up to 80 percent latency reductio…
NVIDIA announced that DFlash speculative decoding boosts inference performance on Blackwell GPUs by up to 15x for gpt-oss-120b and nearly doubles interactivity for Llama 3.1 8B compared to EAGLE-3. Th…
Diffusion Language Models (dLLMs) offer a faster alternative to autoregressive LLMs by generating multiple tokens in parallel, but face consistency challenges. Recent innovations like DFlash, Diffusio…
Researchers have developed adaptive speculative decoding, a method that dynamically selects draft lengths at runtime to optimize token generation efficiency in large language models. The approach addr…
Modal Labs released state-of-the-art DFlash speculators for Qwen 3.5 and Qwen 3.6 models on Hugging Face, achieving 5-20% additional speedups and enabling Qwen 3.5 122B-A10B to run at over 1000 tok/s …
Jian Chen, Yesheng Liang, and Zhijian Liu integrated Z Lab's DFlash block diffusion speculative decoding method into SGLang, collaborating with Modal and LMSYS. The team reports up to 4.31x higher thr…
Xiaomi's MiMo-V2.5-Pro-UltraSpeed, a 1.02-trillion-parameter MoE model, achieved 1,000 tokens per second inference on standard cloud GPUs using FP4 quantization, DFlash speculative decoding, and TileR…
Speculative decoding, a lossless inference optimisation that predicts future tokens to reduce latency, faces new economic constraints as modern mixture-of-experts (MoE) architectures replace dense tra…
Researchers at UC San Diego, led by Hao Zhang, have successfully implemented a diffusion-style speculative decoding method called DFlash on Google TPUs, achieving an average 3.13x increase in tokens p…