Metraipolis › archive › dc_fc99f7eefa86
M2Tok: Multi-head Multi-codebook Discrete Action Tokenization for Vision-Language-Action Models — arXiv:2609.18259
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.18259
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T13:49:54.715Z
- SHA-256 of the archived text
94589ee54a109b4b018a5d45b83c34195c9cb05b0fcb3809c55d212e067695d0- Stable URL of this page
https://metraipolis.com/documento/dc_fc99f7eefa86
How to cite
M2Tok: Multi-head Multi-codebook Discrete Action Tokenization for Vision-Language-Action Models — arXiv:2609.18259 — arXiv (en), https://arxiv.org/abs/2609.18259 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fc99f7eefa86 (Metraipolis archive dc_fc99f7eefa86, sha256 94589ee54a109b4b018a5d45b83c34195c9cb05b0fcb3809c55d212e067695d0)
BibTeX
@misc{dcfc99f7eefa86,
title = {M2Tok: Multi-head Multi-codebook Discrete Action Tokenization for Vision-Language-Action Models — arXiv:2609.18259},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.18259},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fc99f7eefa86; sha256 94589ee54a109b4b018a5d45b83c34195c9cb05b0fcb3809c55d212e067695d0},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.18259. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Recent advancements have successfully adapted autoregressive language models to process multimodal signals, such as images and actions. Since raw action signals are continuous, effective tokenization is essential to map high-dimensional inputs into compact discrete tokens for autoregressive processing. However, existing discrete action tokenizers often suffer from high reconstruction loss, failing to preserve the fine-grained dynamics required for precise control. This "discretization bottleneck" significantly limits the performance ceiling of downstream Vision-Language-Action (VLA) models. To address this, we propose ${M}^2$Tok, a Multi-head Multi-codebook Action Tokenizer designed to minimize reconstruction error and enhance policy performance. Our approach introduces two key structural innovations: (1) we decompose the latent action features into multiple heads, enabling the model to implicitly align specific heads with distinct action dimensions; (2) we assign independent codebooks to each head for quantization. By leveraging the combinatorial nature of multiple codebooks, we significantly expand the representational expressivity of the tokenizer, leading to substantially lower reconstruction loss compared to previous methods. We evaluate the ${M}^2$Tok-based VLA on the RoboTwin, Simpler-Env, and 3 zero-shot real-world tasks. Experimental results demonstrate our method not only achieves superior reconstruction fidelity but also significantly boosts the success rate of VLA models. Comprehensive ablation studies further confirm the effectiveness of the multi-head and multi-codebook mechanisms. Code is available at this https URL .
Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-17 in cs.CL).
Categorie arXiv: Robotics (cs.RO) ; Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.18259 — 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 2026-10-15T13:48:24.411Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-17?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).