AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Efficient-DLM: From Autoregressive to Diffusion Language Models, and Beyond in Speed

Yonggan Fu, Lexington Whalen, Zhifan Ye, Xin Dong, Shizhe Diao, Jingyu Liu, Chengyue Wu, Hao Zhang, Enze Xie, Song Han, Maksim Khadkevich, Jan Kautz and 2 more

ICML 2026 · Architectures and efficiency › Attention and transformers · approach: Sparse attention for long-context LLMs · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use block-wise attention instead of bidirectional (no clean context) (ablation)AR-to-dLM conversion · Qwen2.5 1.5B, 50B tokens, 6 generation tasksaverage accuracy: 19.29 -> 28.23Improved
add clean context to block-wise attention (ablation)AR-to-dLM conversion · Qwen2.5 1.5B, 50B tokens, 6 generation tasksaverage accuracy: 28.23 -> 37.69Improved
remove token shift (ablation)AR-to-dLM conversion · Qwen2.5 1.5B, 50B tokens, 6 generation tasksaverage accuracy: 37.69 -> 38.41Improved
use full best scheme instead of bidirectional attention (ablation)AR-to-dLM conversion · Qwen2.5 1.5B, 50B tokens, 6 generation tasksaverage accuracy: 19.29 -> 38.41Improved
use LoRA rank 64 instead of full-model training (ablation)AR-to-dLM conversion · Qwen2.5 1.5B, 50B tokens, 6 generation tasksaverage accuracy: 38.41 -> 30.78Hurt
increase LoRA rank from 16 to 64 (ablation)AR-to-dLM conversion · Qwen2.5 1.5B, 50B tokens, 6 generation tasksaverage accuracy: 27.93 -> 30.78Improved
add position-dependent masking (ablation)AR-to-dLM conversion · Qwen3 4B, AR-to-dLM components (6 math/coding tasks)average accuracy: 60.27 -> 62.02Improved
scale training to 300B tokens (ablation)AR-to-dLM conversion · Qwen3 4B, AR-to-dLM components (6 math/coding tasks)average accuracy: 62.02 -> 64.05Improved
add block-wise attention with clean context to Dream's bidirectional setting (ablation)AR-to-dLM conversion · Qwen3 4B, AR-to-dLM components (6 math/coding tasks)average accuracy: 44.59 -> 59.01Improved
lower initial learning rate from 1e-4 to 1e-5 (ablation)AR-to-dLM conversion · Qwen3 4B, 25B tokens, initial learning rate sweepaverage accuracy: 50.95 -> 60.63Improved
lower initial learning rate from 1e-5 to 1e-6 (ablation)AR-to-dLM conversion · Qwen3 4B, 25B tokens, initial learning rate sweepaverage accuracy: 60.63 -> 54.56Hurt
use Efficient-DLM 1.5B instead of Qwen2.5 1.5BEfficient-DLM · 12-task average (coding, math, MMLU, commonsense)average accuracy: 54.47 -> 52.04Mixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy