无论我使用多少个周期或是更改学习率，我的验证准确率始终停留在50%左右。目前我使用了一个dropout层，如果我使用两个dropout层，我的最高训练准确率为40%，验证准确率为59%。而现在使用一个dropout层，我的结果如下：
2527/2527 [==============================] - 26s 10ms/step - loss: 1.2076 - accuracy: 0.7944 - val_loss: 3.0905 - val_accuracy: 0.5822Epoch 10/202527/2527 [==============================] - 26s 10ms/step - loss: 1.1592 - accuracy: 0.7991 - val_loss: 3.0318 - val_accuracy: 0.5864Epoch 11/202527/2527 [==============================] - 26s 10ms/step - loss: 1.1143 - accuracy: 0.8034 - val_loss: 3.0511 - val_accuracy: 0.5866Epoch 12/202527/2527 [==============================] - 26s 10ms/step - loss: 1.0686 - accuracy: 0.8079 - val_loss: 3.0169 - val_accuracy: 0.5872Epoch 13/202527/2527 [==============================] - 31s 12ms/step - loss: 1.0251 - accuracy: 0.8126 - val_loss: 3.0173 - val_accuracy: 0.5895Epoch 14/2527/2527 [==============================] - 26s 10ms/step - loss: 0.9824 - accuracy: 0.8165 - val_loss: 3.0013 - val_accuracy: 0.5917Epoch 15/202527/2527 [==============================] - 26s 10ms/step - loss: 0.9417 - accuracy: 0.8216 - val_loss: 2.9909 - val_accuracy: 0.5938Epoch 16/202527/2527 [==============================] - 26s 10ms/step - loss: 0.9000 - accuracy: 0.8264 - val_loss: 3.0269 - val_accuracy: 0.5943Epoch 17/202527/2527 [==============================] - 26s 10ms/step - loss: 0.8584 - accuracy: 0.8332 - val_loss: 3.0011 - val_accuracy: 0.5934Epoch 18/202527/2527 [==============================] - 26s 10ms/step - loss: 0.8172 - accuracy: 0.8378 - val_loss: 2.9918 - val_accuracy: 0.5949Epoch 19/202527/2527 [==============================] - 26s 10ms/step - loss: 0.7796 - accuracy: 0.8445 - val_loss: 2.9974 - val_accuracy: 0.5929Epoch 20/202527/2527 [==============================] - 25s 10ms/step - loss: 0.7407 - accuracy: 0.8502 - val_loss: 3.0005 - val_accuracy: 0.5907最高只能达到59%。这是我得到的图表：
无论我做了多少更改，验证准确率最高只能达到59%。这是我的代码：
BATCH_SIZE = 64EPOCHS = 20LSTM_NODES = 256NUM_SENTENCES = 3000MAX_SENTENCE_LENGTH = 50MAX_NUM_WORDS = 5000EMBEDDING_SIZE = 100encoder_inputs_placeholder = Input(shape=(max_input_len,))x = embedding_layer(encoder_inputs_placeholder)encoder = LSTM(LSTM_NODES, return_state=True)encoder_outputs, h, c = encoder(x)encoder_states = [h, c]decoder_inputs_placeholder = Input(shape=(max_out_len,))decoder_embedding = Embedding(num_words_output, LSTM_NODES)decoder_inputs_x = decoder_embedding(decoder_inputs_placeholder)decoder_lstm = LSTM(LSTM_NODES, return_sequences=True, return_state=True)decoder_outputs, _, _ = decoder_lstm(decoder_inputs_x, initial_state=encoder_states)decoder_dropout1 = Dropout(0.2)decoder_outputs = decoder_dropout1(decoder_outputs)decoder_dense1 = Dense(num_words_output, activation='softmax')decoder_outputs = decoder_dense1(decoder_outputs)opt = tf.keras.optimizers.RMSprop()model = Model([encoder_inputs_placeholder, decoder_inputs_placeholder], decoder_outputs)model.compile( optimizer=opt, loss='categorical_crossentropy', metrics=['accuracy'])history = model.fit( [encoder_input_sequences, decoder_input_sequences], decoder_targets_one_hot, batch_size=BATCH_SIZE, epochs=EPOCHS, validation_split=0.1,)我非常困惑为什么只有我的训练准确率在更新，验证准确率却没有变化。
这是模型摘要：
Model: "model_1"__________________________________________________________________________________________________Layer (type) Output Shape Param # Connected to ==================================================================================================input_1 (InputLayer) (None, 25) 0 __________________________________________________________________________________________________input_2 (InputLayer) (None, 23) 0 __________________________________________________________________________________________________embedding_1 (Embedding) (None, 25, 100) 299100 input_1[0][0] __________________________________________________________________________________________________embedding_2 (Embedding) (None, 23, 256) 838144 input_2[0][0] __________________________________________________________________________________________________lstm_1 (LSTM) [(None, 256), (None, 365568 embedding_1[0][0] __________________________________________________________________________________________________lstm_2 (LSTM) [(None, 23, 256), (N 525312 embedding_2[0][0] lstm_1[0][1] lstm_1[0][2] __________________________________________________________________________________________________dropout_1 (Dropout) (None, 23, 256) 0 lstm_2[0][0] __________________________________________________________________________________________________dense_1 (Dense) (None, 23, 3274) 841418 dropout_1[0][0] ==================================================================================================Total params: 2,869,542Trainable params: 2,869,542Non-trainable params: 0__________________________________________________________________________________________________None回答：
训练数据集的大小不到3K，而可训练参数的数量约为300万。你问题的答案是经典的过拟合——模型太大，只是记住了训练子集而不是进行泛化。
如何改善当前情况：
