资讯

腾讯AI Lab首创RL框架Parallel-R1，教大模型学会「并行思维」

在这项工作中，研究者们提出了 Parallel-R1，这是首个能在真实的通用数学推理任务上，通过强化学习教会大模型进行并行思维的框架。除此之外，研究者们进一步对并行思考行为以及其潜在价值进行了深入探讨。 返回搜狐，查看更多 ...

Tsinghua Team Disrupts Traditional Thinking, Introducing a New Paradigm of Native Parallel ...

In this context, a recent study by the Tsinghua University Institute of Artificial Intelligence Research (AIR) has attracted widespread attention. The research paper titled "ParaThinker: Native ...