关于训练集和验证集loss损失函数过大的问题 指令为history = model.fit(x=x_train,y=y_train,batch_size=2,epochs=100,validation_split=0.2)

结果为 Epoch 1/100 3/3 [==============================] - 1s 106ms/step - loss: 31754298.0000 - val_loss: 276263272448.0000 Epoch 2/100 3/3 [==============================] - 0s 16ms/step - loss: 19134654829297664.0000 - val_loss: 36587841507083419648.0000 Epoch 3/100 3/3 [==============================] - 0s 14ms/step - loss: 573201443209899931598848.0000 - val_loss: 51133710489536223535693824.0000 Epoch 4/100 3/3 [==============================] - 0s 13ms/step - loss: 2634256312263739755802512064512.0000 - val_loss: 22880625392784690823271580988080128.0000 Epoch 5/100 3/3 [==============================] - 0s 14ms/step - loss: inf - val_loss: inf

。。。 损失函数过大，第5轮以后就没有数据了。