arXiv cs.CL· Guanghao Zhu, Zeyu Liu, Zhitian Hou, Pengkai Wang, Zhijie Sang, Shuo Cai, Yang Yu, Yuanyi Wang, Yanggan Gu, Congkai Xie, Jianmin Wu, Hongxia Yang·· 10 天前AI 评分35
InfiMed2：基于上下文证据与稳定性感知监督的通用医学多模态基础模型
InfiMed2: A Generalist Medical Multimodal Foundation Model from Contextual Evidence and Stability-Aware Supervision
AI 导读
InfiMed2 是一个 4B 和 27B 的通用医学多模态基础模型系列，基于 55.68B token 语料和阶段感知数据设计构建。其 CPT 流程先适配视觉编码器、再建立广泛医学知识、最后在学习率衰减阶段转向证据聚焦数据混合；SFT 阶段通过答案稳定性、答案掩码重建和正确性约束筛选重新生成视觉问答回复。
正文 · 原文
Authors:Guanghao Zhu, Zeyu Liu, Zhitian Hou, Pengkai Wang, Zhijie Sang, Shuo Cai, Yang Yu, Yuanyi Wang, Yanggan Gu, Congkai Xie, Jianmin Wu, Hongxia Yang
Abstract:Recent medical multimodal models have benefited from larger corpora, broader modality coverage, and stronger reasoning-oriented training, yet effective data design across continued pretraining (CPT) and post-training remains challenging. Medical sources vary substantially in structure, granularity, and information density, and their utility shifts as training progresses from broad knowledge acquisition to late-stage consolidation. Meanwhile, post-training is often dominated by short-form visual question answering, providing limited supervision for informative and answer-consistent explanations. We introduce InfiMed2, a family of 4B and 27B generalist medical multimodal foundation models built around stage-aware data design. We curate a 55.68B-token corpus that combines broad clinical knowledge with context-rich biomedical visual evidence through source-specific processing. Our CPT pipeline first adapts the vision encoder, then builds broad medical knowledge, and finally transitions to an evidence-focused data mixture during learning-rate decay. For supervised fine-tuning (SFT), we regenerate visual question-answering responses using answer stability, answer-masked reconstruction, and correctness-constrained selection to produce more informative and answer-consistent supervision. The 4B model is further optimized with reinforcement learning with verifiable rewards (RLVR). Across five medical multimodal benchmarks, InfiMed2-4B achieves 66.73% mean accuracy after RLVR, surpassing the larger Qwen3.5-9B, while InfiMed2-27B reaches 73.72%, the highest among the evaluated open-weight models.
| Subjects: | Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2609.34798 [cs.CL] |
| (or arXiv:2609.34798v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2609.34798 arXiv-issued DOI via DataCite |
Submission history
From: Guanghao Zhu [view email]
[v1]
Mon, 28 Sep 2026 10:01:35 UTC (1,334 KB)
[v2]
Wed, 7 Oct 2026 03:24:07 UTC (1,334 KB)
来源：arXiv cs.CL · arxiv.org