C-RADIOv4通过多教师蒸馏，提升视觉模型性能与灵活性。
C-RADIOv4 (Tech Report)
- 采用多教师蒸馏融合SigLIP2、DINOv3、SAM3优势
- 在相同算力下，下游任务表现显著提升，支持任意分辨率
- 新增ViTDet选项，高分辨率下效率大幅优化，开源可商用
本文介绍C-RADIO系列最新版本C-RADIOv4，基于AM-RADIO/RADIOv2.5设计，通过多教师蒸馏构建统一学生模型，保留并增强多个教师的独特能力。新版本释放- SO400M（412M参数）和-H（631M）两种变体，训练使用更新的教师模型：SigLIP2、DINOv3和SAM3。相比前代，在核心指标上实现显著提升，并因模仿SAM3引入新能力。此外，进一步强化任意分辨率支持，重新加入ViTDet选项，大幅提高高分辨率推理效率，且采用宽松许可协议，便于广泛应用。
原文摘要 · Abstract (English)
By leveraging multi-teacher distillation, agglomerative vision backbones provide a unified student model that retains and improves the distinct capabilities of multiple teachers. In this tech report, we describe the most recent release of the C-RADIO family of models, C-RADIOv4, which builds upon AM-RADIO/RADIOv2.5 in design, offering strong improvements on key downstream tasks at the same computational complexity. We release -SO400M (412M params), and -H (631M) model variants, both trained with an updated set of teachers: SigLIP2, DINOv3, and SAM3. In addition to improvements on core metrics and new capabilities from imitating SAM3, the C-RADIOv4 model family further improves any-resolution support, brings back the ViTDet option for drastically enhanced efficiency at high-resolution, and comes with a permissive license.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。