SPA-Cache: Singular Proxies for Adaptive Caching in Diffusion Language Models
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866916041998729216 |
|---|---|
| author | Sun, Wenhao Tu, Rong-Cheng Ding, Yifu Jin, Zhao Liao, Jingyi Jing, Yongcheng Tao, Dacheng |
| author_facet | Sun, Wenhao Tu, Rong-Cheng Ding, Yifu Jin, Zhao Liao, Jingyi Jing, Yongcheng Tao, Dacheng |
| contents | While Diffusion Language Models (DLMs) offer a flexible, arbitrary-order alternative to the autoregressive paradigm, their non-causal nature precludes standard KV caching, forcing costly hidden state recomputation at every decoding step. Existing DLM caching approaches reduce this cost by selective hidden state updates; however, they are still limited by (i) costly token-wise update identification heuristics and (ii) rigid, uniform budget allocation that fails to account for heterogeneous hidden state dynamics. To address these challenges, we present SPA-Cache that jointly optimizes update identification and budget allocation in DLM cache. First, we derive a low-dimensional singular proxy that enables the identification of update-critical tokens in a low-dimensional subspace, substantially reducing the overhead of update identification. Second, we introduce an adaptive strategy that allocates fewer updates to stable layers without degrading generation quality. Together, these contributions significantly improve the efficiency of DLMs, yielding up to an $8\times$ throughput improvement over vanilla decoding and a $2$--$4\times$ speedup over existing caching baselines. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_02544 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | SPA-Cache: Singular Proxies for Adaptive Caching in Diffusion Language Models Sun, Wenhao Tu, Rong-Cheng Ding, Yifu Jin, Zhao Liao, Jingyi Jing, Yongcheng Tao, Dacheng Machine Learning Artificial Intelligence While Diffusion Language Models (DLMs) offer a flexible, arbitrary-order alternative to the autoregressive paradigm, their non-causal nature precludes standard KV caching, forcing costly hidden state recomputation at every decoding step. Existing DLM caching approaches reduce this cost by selective hidden state updates; however, they are still limited by (i) costly token-wise update identification heuristics and (ii) rigid, uniform budget allocation that fails to account for heterogeneous hidden state dynamics. To address these challenges, we present SPA-Cache that jointly optimizes update identification and budget allocation in DLM cache. First, we derive a low-dimensional singular proxy that enables the identification of update-critical tokens in a low-dimensional subspace, substantially reducing the overhead of update identification. Second, we introduce an adaptive strategy that allocates fewer updates to stable layers without degrading generation quality. Together, these contributions significantly improve the efficiency of DLMs, yielding up to an $8\times$ throughput improvement over vanilla decoding and a $2$--$4\times$ speedup over existing caching baselines. |
| title | SPA-Cache: Singular Proxies for Adaptive Caching in Diffusion Language Models |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2602.02544 |