One-Step Text Generation by Seq-Drifting
We propose Seq-Drifting, a framework that applies drifting to sequences of continuous token representations and directly learns a map from Gaussian noise to text.
ProofPaper ↗
Key points
- Diffusion language models generate tokens in parallel, but still require multiple sampling steps to produce high-quality text.
- In this paper, we investigate whether a model can learn to generate an entire sequence in a single forward pass.
- Unlike existing DLMs that derive the supervision at each position directly from the observed token in a training sequence, Seq-Drifting builds a drift field whose position-wise targets adapt to the current generated context.
- These results suggest that drifting over token sequences offers a promising path toward effective one-step text generation.
Sources (1)
- [1]One-Step Text Generation by Seq-DriftingarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 9, 02:15 AM
We propose Seq-Drifting, a framework that applies drifting to sequences of continuous token representations and directly learns a map from Gaussian noise to text.
Diffusion language models generate tokens in parallel, but still require multiple sampling steps to produce high-quality text.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 9, 2026Work2Skill: Capturing Human GUI Experience for Reusable Agent Skill Induction
- Oct 9, 2026AMLworld-Compact: Importance-Weighted Downsampling for Cost-Effective LLM Evaluation and Error Diagnosis
- Oct 8, 2026openai/codex rust-v0.162.0: 0.162.0
- Oct 8, 2026Microsoft Joins the Local AI Push
- Oct 8, 2026Understanding Jev, the new model everyone is talking about
- Oct 7, 2026[AINews] Claude Haiku 5.5 — better than GPT-6 Luna at the same pricing