pytorch的HingeEmbeddingLoss 有先后順序
L1Loss MSELoss 都沒問題,但是 HingeEmbeddingLoss 卻總報錯 說不能求梯度 #criterion = nn.MSELoss()#criterion = nn. ...
L1Loss MSELoss 都沒問題,但是 HingeEmbeddingLoss 卻總報錯 說不能求梯度 #criterion = nn.MSELoss()#criterion = nn. ...