第二块是M层的cross-attention和Multi-Layer Perceptron块。MLO视图和CC视图经过交叉注意力层的输出总结可以写成：翻译成英文
The second block consists of the cross-attention and Multi-Layer Perceptron (MLP) blocks in the M layer. The summarized outputs of the MLO and CC views after passing through the cross-attention layer can be expressed as:
原文地址: https://www.cveoy.top/t/topic/fUuX 著作权归作者所有。请勿转载和采集!