cd/entity/Diffusion Large Language Models· home› entities› Diffusion Large Language Models

grep -l @diffusion large language models /news/*.json | wc -l → 2

Diffusion Large Language Models

mentions 2 type Person feed RSS

// recent coverage 2 mentions

04:00

2026-07-24

arxiv.org

large-language-models

DC-Leap: Training-Free Acceleration of dLLMs via Draft-Guided Contiguous Leaping Decoding

Researchers propose DC-Leap, a training-free framework that accelerates Diffusion Large Language Models (dLLMs) by up to 53.19x on MBPP for long-sequence generation and up to 105.02x when combined wit…

04:00

2026-06-19

arxiv.org

large-language-models

Where to Place the Query? Unveiling and Mitigating Positional Bias in In-Context Learning for Diffusion LLMs via Decoding Dynamics

Researchers have identified and characterized positional bias in in-context learning for diffusion large language models (dLLMs), showing that query placement significantly affects generation quality.…

// co-occurs with top 5 entities

Auto-ICL 1 arXiv 1 DC-Leap 1 MBPP 1 KV-Cache 1

// topics top 6 topics

large language models 2 machine learning 1 natural language processing 1 ai research 1 artificial intelligence 1 generative ai 1