第三块是M层的self-attention和cross-attention和Multi-Layer Perceptron块。MLO视图和CC视图经过全注意力层的输出总结可以写成：翻译成英文
The third block consists of the self-attention, cross-attention, and Multi-Layer Perceptron (MLP) blocks of the M layer. The outputs of the MLO view and CC view after the fully-attention layer can be summarized as follows:
原文地址: https://www.cveoy.top/t/topic/fUvJ 著作权归作者所有。请勿转载和采集!