dorsal/arxiv
View SchemaBeyond Hard Masks: Progressive Token Evolution for Diffusion Language Models
| Authors | Linhao Zhong, Linyu Wu, Bozhen Fang, Tianjian Feng, Chenchen Jing, Wen Wang, Jiaheng Zhang, Hao Chen, Chunhua Shen |
|---|---|
| Categories | |
| ArXiv ID | 2601.07351vv1 |
| URL | https://arxiv.org/abs/2601.07351 |
| License | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ |
Abstract
Diffusion Language Models (DLMs) offer a promising alternative for language modeling by enabling parallel decoding through iterative refinement. However, most DLMs rely on hard binary masking and discrete token assignments, which hinder the revision of early decisions and underutilize intermediate probabilistic representations. In this paper, we propose EvoToken-DLM, a novel diffusion-based language modeling approach that replaces hard binary masks with evolving soft token distributions. EvoToken-DLM enables a progressive transition from masked states to discrete outputs, supporting revisable decoding. To effectively support this evolution, we introduce continuous trajectory supervision, which aligns training objectives with iterative probabilistic updates. Extensive experiments across multiple benchmarks show that EvoToken-DLM consistently achieves superior performance, outperforming strong diffusion-based and masked DLM baselines. Project webpage: https://aim-uofa.github.io/EvoTokenDLM.
{
"annotation_id": "2a1d9694-e191-4677-af66-ac3027cea2da",
"date_created": "2026-02-17T05:53:12.485000Z",
"date_modified": "2026-02-17T05:53:12.485000Z",
"file_hash": "b59b29487cc602dd8437f9d8480c2cba3c3866fdd5604a447a8e7de63c520ba7",
"private": false,
"record": {
"abstract": "Diffusion Language Models (DLMs) offer a promising alternative for language modeling by enabling parallel decoding through iterative refinement. However, most DLMs rely on hard binary masking and discrete token assignments, which hinder the revision of early decisions and underutilize intermediate probabilistic representations. In this paper, we propose EvoToken-DLM, a novel diffusion-based language modeling approach that replaces hard binary masks with evolving soft token distributions. EvoToken-DLM enables a progressive transition from masked states to discrete outputs, supporting revisable decoding. To effectively support this evolution, we introduce continuous trajectory supervision, which aligns training objectives with iterative probabilistic updates. Extensive experiments across multiple benchmarks show that EvoToken-DLM consistently achieves superior performance, outperforming strong diffusion-based and masked DLM baselines. Project webpage: https://aim-uofa.github.io/EvoTokenDLM.",
"arxiv_id": "2601.07351",
"authors": [
"Linhao Zhong",
"Linyu Wu",
"Bozhen Fang",
"Tianjian Feng",
"Chenchen Jing",
"Wen Wang",
"Jiaheng Zhang",
"Hao Chen",
"Chunhua Shen"
],
"categories": [
"cs.CL",
"cs.AI"
],
"license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
"title": "Beyond Hard Masks: Progressive Token Evolution for Diffusion Language Models",
"url": "https://arxiv.org/abs/2601.07351",
"version": "v1"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "ef8a54f4-0d42-46df-a5fc-1b7c08862c67",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}