Investigating whether text diffusion advantages extend beyond code to mathematical reasoning, structured text, general prose, and translation through bidirectionality analysis, augmentation estimation, and simulated decoding comparison.
Stable-DiffCoder demonstrated that diffusion-based LLMs outperform autoregressive baselines on code generation. The open question: do these benefits extend to other domains? This work answers affirmatively through three complementary analyses across five domains.
Values near 1.0 indicate symmetric forward/backward dependencies. General text and translation show perfect symmetry (beta=1.0). Code and math show slight forward dominance. Structured text has strongest asymmetry (beta=0.926).
| Domain | Mean Length | Unique Tokens | Constraint Density | Eff. Multiplier |
|---|---|---|---|---|
| Code | 24.3 | 124 | 0.104 | 177,169x |
| Math Reasoning | 18.1 | 162 | 0.086 | 5,156x |
| Structured Text | 14.4 | 160 | 0.089 | 562x |
| General Text | 14.4 | 195 | 0.010 | 487x |
| Translation | 11.9 | 153 | 0.034 | 99x |