Mind the Gap No More: Achieving Zero-Gap Multimodal Integration via One Tokenizer
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866913104026140672 |
|---|---|
| author | Li, Yanan Jin, Christina Yi Jin, Yuan Luo, Manli Xu, Tie Jiao, Shuai He, Wei Zhang, Qing |
| author_facet | Li, Yanan Jin, Christina Yi Jin, Yuan Luo, Manli Xu, Tie Jiao, Shuai He, Wei Zhang, Qing |
| contents | A central challenge in developing Multimodal Large Language Models (MLLMs) is effectively integrating heterogeneous inputs into a cohesive reasoning engine. Current paradigms predominantly rely on modular architectures that introduce modality-specific encoders and cross-modal fusion mechanisms. However, these designs are fundamentally bottlenecked by a geometric modality gap, forcing the LLM to expend significant computational capacity on geometric reconciliation rather than deep cross-modal reasoning. In this work, we formally characterize this modality gap and theoretically demonstrate that native architectures, specifically those employing a unified vocabulary, intrinsically maintain a zero-gap state across all hidden layers. Guided by these theoretical findings, we propose \textit{One Tokenizer}, a native architecture that maps all modalities directly into a shared token space. We empirically validate this framework on a DNA--text multimodal testbed. Our extensive evaluations reveal that by achieving seamless integration within the LLM's native latent space, One Tokenizer consistently outperforms encoder-based modular counterparts, providing a fundamentally superior framework for deep biological reasoning. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_12286 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Mind the Gap No More: Achieving Zero-Gap Multimodal Integration via One Tokenizer Li, Yanan Jin, Christina Yi Jin, Yuan Luo, Manli Xu, Tie Jiao, Shuai He, Wei Zhang, Qing Genomics Computation and Language A central challenge in developing Multimodal Large Language Models (MLLMs) is effectively integrating heterogeneous inputs into a cohesive reasoning engine. Current paradigms predominantly rely on modular architectures that introduce modality-specific encoders and cross-modal fusion mechanisms. However, these designs are fundamentally bottlenecked by a geometric modality gap, forcing the LLM to expend significant computational capacity on geometric reconciliation rather than deep cross-modal reasoning. In this work, we formally characterize this modality gap and theoretically demonstrate that native architectures, specifically those employing a unified vocabulary, intrinsically maintain a zero-gap state across all hidden layers. Guided by these theoretical findings, we propose \textit{One Tokenizer}, a native architecture that maps all modalities directly into a shared token space. We empirically validate this framework on a DNA--text multimodal testbed. Our extensive evaluations reveal that by achieving seamless integration within the LLM's native latent space, One Tokenizer consistently outperforms encoder-based modular counterparts, providing a fundamentally superior framework for deep biological reasoning. |
| title | Mind the Gap No More: Achieving Zero-Gap Multimodal Integration via One Tokenizer |
| topic | Genomics Computation and Language |
| url | https://arxiv.org/abs/2602.12286 |