RuntimeError: mat1 dim 1 must match mat2 dim 0 报错解决

原因是输入特征的维度不匹配导致

解决方案1：使用avg_pool2d函数将特征图转换成1维

self.linear3 = nn.Linear(96, 128)

'''
'''
#如果输入的每个特征图的大小是10x2，则用kernel_size=10x2, 对每个10x2的特征图进行avg_pool
out = F.avg_pool2d(out, (10,2))
out = out.view(out.size(0), -1)
out = self.linear3(out)

解决方案2：使用AdaptiveAvgPool2d自适应平均池化，不用额外计算特征图大小

self.linear3 = nn.Linear(96, 128)
self.avg = nn.AdaptiveAvgPool2d(1)#自适应平均池化

'''
'''
out = self.avg(out)
out = out.view(out.size(0), -1)
out = self.linear3(out)