UniICL: Systematizing Unified Multimodal In-context Learning through a Capability-Oriented Taxonomy
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_ | 1866914422603120640 |
|---|---|
| author | Xu, Yicheng Zhang, Jiangning Xue, Zhucun Hu, Teng Yi, Ran Hu, Xiaobin Liu, Yong Tao, Dacheng |
| author_facet | Xu, Yicheng Zhang, Jiangning Xue, Zhucun Hu, Teng Yi, Ran Hu, Xiaobin Liu, Yong Tao, Dacheng |
| contents | In-context Learning enables training-free adaptation via demonstrations but remains highly sensitive to example selection and formatting. In unified multimodal models spanning understanding and generation, this sensitivity is exacerbated by cross-modal interference and varying cognitive demands. Consequently, In-context Learning efficacy is often non-monotonic and highly task-dependent. To diagnose these behaviors, we introduce a six-level capability-oriented taxonomy that categorizes the functional role of demonstrations from basic perception to high-order discernment. Guided by this cognitive framework, we construct UniICL-760K, a large-scale corpus featuring curated 8-shot In-context Learning episodes across 15 subtasks, alongside UniICL-Bench for rigorous, controlled evaluation. As an architectural intervention to stabilize few-shot adaptation, we propose the Context-Adaptive Prototype Modulator, a lightweight, plug-and-play module. Evaluations on UniICL-Bench show that our approach yields highly competitive unified results, outperforming larger-parameter multimodal large language model baselines on most understanding In-context Learning tasks. Data and code will be available soon at https://github.com/xuyicheng-zju/UniICL. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_24690 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | UniICL: Systematizing Unified Multimodal In-context Learning through a Capability-Oriented Taxonomy Xu, Yicheng Zhang, Jiangning Xue, Zhucun Hu, Teng Yi, Ran Hu, Xiaobin Liu, Yong Tao, Dacheng Computer Vision and Pattern Recognition In-context Learning enables training-free adaptation via demonstrations but remains highly sensitive to example selection and formatting. In unified multimodal models spanning understanding and generation, this sensitivity is exacerbated by cross-modal interference and varying cognitive demands. Consequently, In-context Learning efficacy is often non-monotonic and highly task-dependent. To diagnose these behaviors, we introduce a six-level capability-oriented taxonomy that categorizes the functional role of demonstrations from basic perception to high-order discernment. Guided by this cognitive framework, we construct UniICL-760K, a large-scale corpus featuring curated 8-shot In-context Learning episodes across 15 subtasks, alongside UniICL-Bench for rigorous, controlled evaluation. As an architectural intervention to stabilize few-shot adaptation, we propose the Context-Adaptive Prototype Modulator, a lightweight, plug-and-play module. Evaluations on UniICL-Bench show that our approach yields highly competitive unified results, outperforming larger-parameter multimodal large language model baselines on most understanding In-context Learning tasks. Data and code will be available soon at https://github.com/xuyicheng-zju/UniICL. |
| title | UniICL: Systematizing Unified Multimodal In-context Learning through a Capability-Oriented Taxonomy |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2603.24690 |