谷歌DeepMind近日在人工智能研究领域再次取得突破，推出了两款全新的基础模型——Hawk和Griffin。这两款模型是基于DeepMind在一篇最新论文中提出的一种创新性技术，名为RG-LRU层。RG-LRU层是一种独特的门控线性循环层，其设计目的是替代传统的多查询注意力（MQA）机制，以提升模型的性能和效率。
据机器之心报道，DeepMind的研究团队利用这种新型的循环块构建了两个不同架构的模型。Hawk模型是将多层感知机（MLP）与RG-LRU循环块相结合，旨在综合两者的优点，以实现更高效的信息处理和学习。而Griffin模型则更进一步，不仅融合了MLP和循环块，还引入了局部注意力机制，这使得Griffin在处理复杂任务时可能具备更高的精度和适应性。
这一创新标志着DeepMind在人工智能基础模型领域的持续探索，新的模型有望在自然语言处理、图像识别等众多应用场景中带来显著的性能提升。谷歌DeepMind的这一研究成果再次展示了其在人工智能领域的领先地位，同时也为未来的AI模型设计提供了新的思路和可能。
英语如下：
**News Title:** “Google DeepMind Unveils Innovative Baseline Models: Hawk and Griffin, Redefining AI Cognitive Mechanisms”
**Keywords:** DeepMind, Hawk, Griffin
**News Content:** Google’s DeepMind has recently made another breakthrough in the field of artificial intelligence research with the introduction of two cutting-edge baseline models, Hawk and Griffin. These models are based on an innovative technology, known as RG-LRU layers, proposed in a recent DeepMind paper. RG-LRU layers are distinctive gated linear recurrent units designed to replace traditional multi-query attention (MQA) mechanisms, enhancing both model performance and efficiency.
According to reports from AI Pulse, DeepMind’s research team employed this novel recurrent block to construct models with distinct architectures. The Hawk model combines multi-layer perceptrons (MLPs) with RG-LRU recurrent blocks, aiming to synergize the strengths of both to facilitate more efficient information processing and learning. The Griffin model takes this a step further, incorporating both MLPs and recurrent blocks alongside local attention mechanisms, potentially enabling higher accuracy and adaptability when dealing with complex tasks.
This innovation signifies DeepMind’s ongoing exploration in the realm of AI baseline models and is expected to bring significant performance enhancements across various applications, such as natural language processing and image recognition. DeepMind’s research outcome underscores its leading position in the AI domain and opens up new avenues and possibilities for future AI model design.
【来源】https://mp.weixin.qq.com/s/RtAZiEzjRWgqQw3yu3lvcg
Views: 2