• 医疗多模态模型开始超越GPT-4o级闭源系统，MedXIAOHE用实体感知预训练+RL推理训练打通了从罕见病到长报告生成的全链路

• 小米开源机器人VLA模型，消费级GPU上实现实时双臂操控，从训练到部署的异步执行设计是关键

• 科学工具调用是Agent的硬伤。SciAgentGym造了1780个领域工具做压测，8B模型微调后反超235B

• RL微调让VLM的benchmark分数上去了，但推理链的"忠实度"却在下降——准确率和可靠性的trade-off浮出水面

阅读全文 →

• A medical multimodal model now outperforms GPT-4o-class closed-source systems. MedXIAOHE chains entity-aware pretraining with RL-based reasoning to cover everything from rare diseases to long-form report generation.

• Xiaomi open-sources a robot VLA model that runs real-time bimanual manipulation on a consumer GPU. The key is asynchronous execution baked into training, not just deployment.

• Scientific tool use is agents' Achilles' heel. SciAgentGym stress-tests 1,780 domain tools — and an 8B fine-tuned model beats a 235B general-purpose one.

• RL fine-tuning boosts VLM benchmark scores, but chain-of-thought faithfulness degrades — surfacing a hidden accuracy-vs-reliability trade-off.

Read more →