老师，下午好！
本节的11：03秒处，可以看到在验证集上准确率是 0.8342，我运行你git上的代码，也是类似的准确率，如下：

history = model.fit(x_train, y_train, epochs = 10, 
                    validation_data = (x_valid, y_valid))

Epoch 1/10
1719/1719 [==============================] - 3s 2ms/step - loss: 8.7104 - accuracy: 0.6663 - val_loss: 0.5905 - val_accuracy: 0.7922
Epoch 2/10
1719/1719 [==============================] - 3s 2ms/step - loss: 0.5476 - accuracy: 0.8003 - val_loss: 0.5241 - val_accuracy: 0.8180
Epoch 3/10
1719/1719 [==============================] - 3s 2ms/step - loss: 0.4774 - accuracy: 0.8255 - val_loss: 0.4888 - val_accuracy: 0.8302
Epoch 4/10
1719/1719 [==============================] - 3s 2ms/step - loss: 0.4387 - accuracy: 0.8398 - val_loss: 0.4630 - val_accuracy: 0.8414
Epoch 5/10
1719/1719 [==============================] - 3s 2ms/step - loss: 0.4131 - accuracy: 0.8456 - val_loss: 0.4496 - val_accuracy: 0.8444
Epoch 6/10
1719/1719 [==============================] - 3s 2ms/step - loss: 0.3961 - accuracy: 0.8544 - val_loss: 0.4409 - val_accuracy: 0.8520
Epoch 7/10
1719/1719 [==============================] - 3s 2ms/step - loss: 0.3865 - accuracy: 0.8549 - val_loss: 0.4524 - val_accuracy: 0.8412
Epoch 8/10
1719/1719 [==============================] - 3s 2ms/step - loss: 0.3702 - accuracy: 0.8622 - val_loss: 0.4279 - val_accuracy: 0.8496
Epoch 9/10
1719/1719 [==============================] - 3s 2ms/step - loss: 0.3677 - accuracy: 0.8648 - val_loss: 0.4320 - val_accuracy: 0.8560
Epoch 10/10
1719/1719 [==============================] - 3s 2ms/step - loss: 0.3516 - accuracy: 0.8694 - val_loss: 0.4243 - val_accuracy: 0.8564

但是，因为看到视频11：03 上的batch 大小是 55000/55000, 而非上面的 1719/1719, 我在了解了相关概念后，把上述代码

history = model.fit(x_train, y_train, epochs = 10, 
                    validation_data = (x_valid, y_valid))

改为了

history = model.fit(x_train, y_train, epochs = 10, batch_size=1,
                    validation_data = (x_valid, y_valid))

其余都不变，但是，再次运行的结果却是

Epoch 1/10
55000/55000 [==============================] - 61s 1ms/step - loss: nan - accuracy: 0.1016 - val_loss: nan - val_accuracy: 0.0914
Epoch 2/10
55000/55000 [==============================] - 61s 1ms/step - loss: nan - accuracy: 0.0994 - val_loss: nan - val_accuracy: 0.0914
Epoch 3/10
55000/55000 [==============================] - 59s 1ms/step - loss: nan - accuracy: 0.1006 - val_loss: nan - val_accuracy: 0.0914
Epoch 4/10
55000/55000 [==============================] - 60s 1ms/step - loss: nan - accuracy: 0.1010 - val_loss: nan - val_accuracy: 0.0914
Epoch 5/10
55000/55000 [==============================] - 62s 1ms/step - loss: nan - accuracy: 0.1005 - val_loss: nan - val_accuracy: 0.0914
Epoch 6/10
55000/55000 [==============================] - 59s 1ms/step - loss: nan - accuracy: 0.1008 - val_loss: nan - val_accuracy: 0.0914
Epoch 7/10
55000/55000 [==============================] - 61s 1ms/step - loss: nan - accuracy: 0.1025 - val_loss: nan - val_accuracy: 0.0914
Epoch 8/10
55000/55000 [==============================] - 60s 1ms/step - loss: nan - accuracy: 0.0984 - val_loss: nan - val_accuracy: 0.0914
Epoch 9/10
55000/55000 [==============================] - 59s 1ms/step - loss: nan - accuracy: 0.1010 - val_loss: nan - val_accuracy: 0.0914
Epoch 10/10
55000/55000 [==============================] - 58s 1ms/step - loss: nan - accuracy: 0.1015 - val_loss: nan - val_accuracy: 0.0914

可以看到，改了之后的准确率连 0.1 都达不到，
请问老师，为什么会出现这种情况？
我需要再次修改哪里才可以回复到以前的准确率呢？
谢谢老师解答！！！

补充一点：因为我刚开始学习，所以我使用的是tf版本是：2.4.1 ，其余的版本号与老师你所使用的类似。