PHOTON: Hierarchical Autoregressive Modeling for Lightspeed and Memory-Efficient Language Generation

Yuma Ichikawa, Naoya Takagi, Takumi Nakagawa, Yuzi Kanazawa, Akira Sakai


Abstract
Transformers operate as horizontal token-by-token scanners; at each generation step, attending to an ever-growing sequence of token-level states. This access pattern increases prefill latency and makes long-context decoding more memory-bound, as KV-cache reads and writes dominate inference time over arithmetic operations. We propose Parallel Hierarchical Operation for TOp-down Networks (PHOTON), a hierarchical autoregressive model that replaces horizontal scanning with vertical, multi-resolution context scanning. PHOTON maintains a hierarchy of latent streams: a bottom-up encoder compresses tokens into low-rate contextual states, while lightweight top-down decoders reconstruct fine-grained token representations in parallel. We further introduce recursive generation that updates only the coarsest latent stream and eliminates bottom-up re-encoding. Experimental results show that PHOTON is superior to competitive Transformer-based language models regarding the throughput-quality trade-off, providing advantages in long-context and multi-query tasks. In particular, this reduces decode-time KV-cache traffic, yielding up to 103 × higher throughput per unit memory.
Anthology ID:
2026.acl-long.1778
Volume:
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
Month:
July
Year:
2026
Address:
San Diego, California, United States
Editors:
Maria Liakata, Viviane P. Moreira, Jiajun Zhang, David Jurgens
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
38365–38383
Language:
URL:
https://aclanthology.org/2026.acl-long.1778/
DOI:
10.18653/v1/2026.acl-long.1778
Bibkey:
Cite (ACL):
Yuma Ichikawa, Naoya Takagi, Takumi Nakagawa, Yuzi Kanazawa, and Akira Sakai. 2026. PHOTON: Hierarchical Autoregressive Modeling for Lightspeed and Memory-Efficient Language Generation. In Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 38365–38383, San Diego, California, United States. Association for Computational Linguistics.
Cite (Informal):
PHOTON: Hierarchical Autoregressive Modeling for Lightspeed and Memory-Efficient Language Generation (Ichikawa et al., ACL 2026)
Copy Citation:
PDF:
https://aclanthology.org/2026.acl-long.1778.pdf
Checklist:
 2026.acl-long.1778.checklist.pdf