Research map / Image generation / Fast diffusion sampling
Beyond Scattered Acceptance: Fast and Coherent Inference for DLMs via Longest Stable Prefixes
Pengxiang Li, Jiayin Cai, Hongwei Xue, Kunyu Shi, Shilin Yan
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use adaptive prefix sizing instead of fixed 8-token commits (ablation) | LSP · GSM8K (LLaDA-8B) | score (%): 19.3 -> 69.9 | Improved |
| use adaptive sizing instead of fixed 1-token commits (ablation) | LSP · GSM8K (LLaDA-8B) | total steps: 128 -> ~68 steps, 67.1 -> 69.9 score | Improved |
| add structural snapping to delimiters (ablation) | LSP · GSM8K (LLaDA-8B) | score (%): 67.8 -> 69.9 (steps ~50 -> ~68) | Improved |
| use prefix-first instead of scattered acceptance (ablation) | LSP · GSM8K (LLaDA-8B) | score (%): 68.9 -> 69.9 (steps 128 -> ~68) | Improved |
| use prefix-first instead of scattered acceptance (ablation) | LSP · mid-generation flip rate | flip rate (%): 14.2% -> 4.3% | Improved |
| apply LSP scheduler vs Full decoding | LLaDA-8B · MMLU / HumanEval / GSM8K | score (%) and speedup: GSM8K 77.1 -> 77.6 at 1.51x; HumanEval 30.5 -> 29.3 at 1.22x | No clear effect |
| apply LSP scheduler vs Full decoding | LLaDA-8B · TruthfulQA | score (%): 34.4 -> 45.8 (2.29x) | Improved |
| apply LSP scheduler vs Full decoding | LLaDA-8B · WinoGrande | score (%): 73.8 -> 70.7 (1.74x) | Hurt |
| apply LSP scheduler vs Full decoding | Dream-7B · Sudoku | speedup: 3.36x speedup, 89.0 -> 88.0 | Improved |
| apply LSP scheduler vs Full decoding | LLaDA-8B · WritingPrompts | coherence (1-5): 4.42 -> 4.38 at 1.82x | No clear effect |
Closest papers by meaning
- Efficient Test-Time Scaling via Hierarchical Search and Self-Verification for Discrete Diffusion Language Models (ICML 2026 · similarity 0.91)
- Polestar: Drift-Aware Cache Calibration and Token Commitment for Efficient Inference of Diffusion LLMs (NeurIPS 2026 · similarity 0.90)
- Accelerating Diffusion Language Models via Structured Suffix Modeling (NeurIPS 2026 · similarity 0.90)
- DAWN: Dependency-Aware Fast Inference for Diffusion LLMs (NeurIPS 2026 · similarity 0.90)
- PACE-dLLM: Elastic Block Decoding via Confidence Cliff Estimation for Diffusion Language Models (NeurIPS 2026 · similarity 0.90)
- Fast-dLLM++: Fr\'{e}chet Profile Decoding for Faster Diffusion LLM Inference (NeurIPS 2026 · similarity 0.90)
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.
