biomedicine
PulseAugur coverage of biomedicine — every cluster mentioning biomedicine across labs, papers, and developer communities, ranked by signal.
2 天有情绪数据
-
GraphRAG通过检索连接知识来减少LLM令牌使用量
使用TigerGraph的GraphRAG方法开发的两个项目展示了其在减少令牌使用量和提高大型语言模型答案质量方面的有效性。这两个系统一个专注于网络安全，另一个专注于生物医学，将GraphRAG与传统的纯LLM和基础RAG方法进行了比较。通过利用知识图谱检索连接的实体和关系，GraphRAG为LLM提供了更集中的上下文，从而在保持准确性的同时降低了成本和延迟。
-
新架构应对LLM的灾难性遗忘问题
研究人员开发了新的架构方法，以解决大型语言模型（LLM）在持续预训练和微调过程中出现的灾难性遗忘问题。其中一种方法TFGN引入了一个叠加层，可以在不改变核心Transformer的情况下实现参数高效更新，在不同领域和模型规模下均能显著保留先前的知识。另一种受生物视觉启发的UAM方法，采用双流架构将语义理解与动作控制分离，在VLA模型训练过程中保持多模态能力。这些进展旨在使模型能够持续学习，而不会降低先前获得的知识的性能。
-
AI models demonstrate physician-like reasoning capabilities, prompting questions about future applications.
A new analysis suggests that artificial intelligence systems are demonstrating the capability to reason in ways similar to human physicians. This development raises questions about the future integration and application…
-
MAMMAL AI architecture accelerates biomedical discovery and drug development
Researchers have introduced MAMMAL, a novel multi-modal architecture designed for biomedical discovery. This system integrates molecular data with language models to accelerate research in areas like drug discovery and …
-
RoseCDL algorithm enhances rare event and anomaly detection in large signals
Researchers have developed RoseCDL, a new Convolutional Dictionary Learning algorithm designed to improve the detection of rare events and anomalies in large datasets. This method enhances robustness by incorporating in…