OpenAI News·· 2021-03-04AI 评分40
CLIP 模型中的多模态神经元发现
Multimodal neurons in artificial neural networks
AI 导读
OpenAI 发现 CLIP 模型中的神经元能够对概念进行多模态识别，无论以文字、符号还是概念形式呈现，这有助于解释 CLIP 在分类视觉概念时的准确性，并推动对模型学习关联和偏差的理解。
原文
We’ve discovered neurons in CLIP that respond to the same concept whether presented literally, symbolically, or conceptually. This may explain CLIP’s accuracy in classifying surprising visual renditions of concepts, and is also an important step toward understanding the associations and biases that CLIP and similar models learn.
来源：OpenAI News · openai.com