Metraipolis › archive › dc_58d2cd7e41b3

Fewer Tokens, More Self-Teaching: On-Policy Self-Distillation for Extreme Visual Token Reduction — arXiv:2609.32353

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.32353
License (as archived): CC0-1.0
Archived: 2026-10-07T21:53:55.339Z
SHA-256 of the archived text: 24e0a68fd8126539c3bad5e64102e1305cef2bbfc170ae7eae9d6d354f2d5847
Stable URL of this page: https://metraipolis.com/documento/dc_58d2cd7e41b3

How to cite

Fewer Tokens, More Self-Teaching: On-Policy Self-Distillation for Extreme Visual Token Reduction — arXiv:2609.32353 — arXiv (en), https://arxiv.org/abs/2609.32353 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_58d2cd7e41b3 (Metraipolis archive dc_58d2cd7e41b3, sha256 24e0a68fd8126539c3bad5e64102e1305cef2bbfc170ae7eae9d6d354f2d5847)

BibTeX

@misc{dc58d2cd7e41b3,
  title = {Fewer Tokens, More Self-Teaching: On-Policy Self-Distillation for Extreme Visual Token Reduction — arXiv:2609.32353},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.32353},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_58d2cd7e41b3; sha256 24e0a68fd8126539c3bad5e64102e1305cef2bbfc170ae7eae9d6d354f2d5847},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32353. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Visual token reduction is an effective way to accelerate multimodal large language models (MLLMs), but performance deteriorates rapidly under extremely low token budgets. Existing work has explored both visual-token selection and training-based adaptation to reduced visual inputs. We take a step further by asking how a heavily compressed MLLM should learn from the states induced by its own generations. This setting naturally calls for on-policy self-distillation: a heavily compressed model is supervised on the states induced by its own generations, while its full-token counterpart serves as an information-rich teacher. Based on this insight, we propose LT-OPD, a training framework for extreme visual-token reduction. The student rolls out responses with only a small fraction of visual tokens, and a frozen full-token copy of the same MLLM provides distributional supervision along these student-generated trajectories. To stabilize on-policy learning when visual evidence is severely limited, we further introduce a budget-level curriculum that progressively decreases the token budget during training. Across nine benchmarks on Qwen3.5-4B, LT-OPD raises average retained performance under 5% visual-token retention from 68.6% to 82.3%, outperforming training-free, training-based, and reinforcement-learning baselines at the same budget. The gains transfer consistently to Qwen3.5-9B, GLM-4.6V-9B, and LLaVA-OV-1.5-4B. LT-OPD also reduces KV-cache usage by 85.2% and prefill FLOPs by 85.4% without additional inference overhead, demonstrating that on-policy learning can substantially recover capabilities lost to extreme visual-token reduction.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.CL).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Computation and Language (cs.CL); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.32353 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T21:27:08.734Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).