Abstract:Continuous diffusion language models lag behind autoregressive transformers, partly because diffusion is applied in spaces poorly suited to language denoising and token recovery. We propose DiHAL, a geometry-guided diffusion-transformer hybrid that asks where diffusion should enter a pretrained transformer. DiHAL scores layers with geometry-based proxies, selects a diffusion-friendly hidden-state interface, and replaces the lower transformer prefix with a diffusion bridge while retaining the upper layers and original LM head. By reconstructing the selected-layer hidden state rather than tokens, DiHAL avoids direct continuous-to-discrete recovery. Experiments on 8B-scale backbones show that the geometry score predicts effective shallow insertion layers under a fixed bridge-training protocol and that hidden-state recovery improves over continuous diffusion baselines in a diagnostic comparison matching the diffusion/recovery training budget. These results suggest that hidden-state geometry helps identify where diffusion-based replacement is feasible inside pretrained language models.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2605.14368 [cs.CL] |
| (or arXiv:2605.14368v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2605.14368 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Injin Kong [view email]
[v1]
Thu, 14 May 2026 04:47:54 UTC (250 KB)