Foreign Exchange RNN Help

Model Training

모델 요약 및 fit 관련 정보

SimpleRNN Model

Summary
_________________________________________________________________ Layer (type) Output Shape Param # ================================================================= simple_rnn (SimpleRNN) (None, 50, 50) 2600 simple_rnn_1 (SimpleRNN) (None, 50) 5050 dense (Dense) (None, 1) 51 ================================================================= Total params: 7701 (30.08 KB) Trainable params: 7701 (30.08 KB) Non-trainable params: 0 (0.00 Byte) ________________________________________________________________
Training setting
  • Callbacks: EarlyStopping(monitor='val_loss', patience=3)

  • Loss: Mean_Squared_Error

  • Optimizer: Adam

  • Batch size: 400

History
RNN history
Epoch 1/50 349/349 [==============================] - 26s 57ms/step - loss: 0.0027 - val_loss: 2.4990e-05 Epoch 2/50 349/349 [==============================] - 20s 57ms/step - loss: 2.0591e-05 - val_loss: 1.6920e-05 Epoch 3/50 349/349 [==============================] - 19s 56ms/step - loss: 1.5860e-05 - val_loss: 1.3386e-05 Epoch 4/50 349/349 [==============================] - 19s 55ms/step - loss: 1.3670e-05 - val_loss: 1.1369e-05 Epoch 5/50 349/349 [==============================] - 20s 56ms/step - loss: 1.2269e-05 - val_loss: 1.2177e-05 Epoch 6/50 349/349 [==============================] - 19s 54ms/step - loss: 2.0865e-05 - val_loss: 8.2757e-05 Epoch 7/50 349/349 [==============================] - 20s 56ms/step - loss: 2.6854e-05 - val_loss: 3.3354e-05

LSTM Model

Summary
_________________________________________________________________ Layer (type) Output Shape Param # ================================================================= lstm (LSTM) (None, 100) 40800 dense (Dense) (None, 1) 101 ================================================================= Total params: 40901 (159.77 KB) Trainable params: 40901 (159.77 KB) Non-trainable params: 0 (0.00 Byte) _________________________________________________________________
Training setting
  • Callbacks: EarlyStopping(monitor='val_loss', patience=3)

  • Loss: Mean_Squared_Error

  • Optimizer: Adam

  • Batch size: 400

History
LSTM history
Epoch 1/50 349/349 [==============================] - 10s 8ms/step - loss: 0.0015 - val_loss: 1.1754e-05 Epoch 2/50 349/349 [==============================] - 2s 7ms/step - loss: 1.1712e-05 - val_loss: 1.1424e-05 Epoch 3/50 349/349 [==============================] - 2s 7ms/step - loss: 1.1356e-05 - val_loss: 1.1008e-05 Epoch 4/50 349/349 [==============================] - 2s 7ms/step - loss: 1.1004e-05 - val_loss: 1.0868e-05 Epoch 5/50 349/349 [==============================] - 2s 7ms/step - loss: 1.0609e-05 - val_loss: 1.0144e-05 Epoch 6/50 349/349 [==============================] - 2s 7ms/step - loss: 1.0311e-05 - val_loss: 9.7962e-06 Epoch 7/50 349/349 [==============================] - 2s 7ms/step - loss: 9.7611e-06 - val_loss: 9.3512e-06 Epoch 8/50 349/349 [==============================] - 2s 7ms/step - loss: 9.7010e-06 - val_loss: 9.0441e-06 Epoch 9/50 349/349 [==============================] - 2s 7ms/step - loss: 9.4313e-06 - val_loss: 9.2712e-06 Epoch 10/50 349/349 [==============================] - 2s 7ms/step - loss: 9.2417e-06 - val_loss: 9.3758e-06 Epoch 11/50 349/349 [==============================] - 2s 7ms/step - loss: 9.3990e-06 - val_loss: 8.1832e-06 Epoch 12/50 349/349 [==============================] - 2s 7ms/step - loss: 9.2997e-06 - val_loss: 7.8138e-06 Epoch 13/50 349/349 [==============================] - 2s 7ms/step - loss: 9.2090e-06 - val_loss: 8.5546e-06 Epoch 14/50 349/349 [==============================] - 2s 7ms/step - loss: 8.7962e-06 - val_loss: 7.3502e-06 Epoch 15/50 349/349 [==============================] - 2s 7ms/step - loss: 8.6043e-06 - val_loss: 7.5583e-06 Epoch 16/50 349/349 [==============================] - 2s 7ms/step - loss: 9.2771e-06 - val_loss: 6.8191e-06 Epoch 17/50 349/349 [==============================] - 2s 7ms/step - loss: 7.9040e-06 - val_loss: 1.4065e-05 Epoch 18/50 349/349 [==============================] - 2s 7ms/step - loss: 8.4011e-06 - val_loss: 7.1161e-06 Epoch 19/50 349/349 [==============================] - 2s 7ms/step - loss: 8.1817e-06 - val_loss: 7.5856e-06

GRU Model

Summary
_________________________________________________________________ Layer (type) Output Shape Param # ================================================================= gru (GRU) (None, 50, 50) 7950 gru_1 (GRU) (None, 50) 15300 dense (Dense) (None, 1) 51 ================================================================= Total params: 23301 (91.02 KB) Trainable params: 23301 (91.02 KB) Non-trainable params: 0 (0.00 Byte) _________________________________________________________________
Training setting
  • Callbacks: EarlyStopping(monitor='val_loss', patience=3)

  • Loss: Mean_Squared_Error

  • Optimizer: Adam

  • Batch size: 400

History
GRU history.png
Epoch 1/50 349/349 [==============================] - 12s 10ms/step - loss: 0.0025 - val_loss: 7.5941e-06 Epoch 2/50 349/349 [==============================] - 3s 8ms/step - loss: 7.0093e-06 - val_loss: 6.1103e-06 Epoch 3/50 349/349 [==============================] - 3s 8ms/step - loss: 5.8589e-06 - val_loss: 5.3030e-06 Epoch 4/50 349/349 [==============================] - 3s 8ms/step - loss: 5.2649e-06 - val_loss: 4.8958e-06 Epoch 5/50 349/349 [==============================] - 3s 8ms/step - loss: 4.9381e-06 - val_loss: 4.6823e-06 Epoch 6/50 349/349 [==============================] - 3s 8ms/step - loss: 4.7218e-06 - val_loss: 4.4238e-06 Epoch 7/50 349/349 [==============================] - 3s 8ms/step - loss: 4.5645e-06 - val_loss: 4.3088e-06 Epoch 8/50 349/349 [==============================] - 3s 8ms/step - loss: 4.4203e-06 - val_loss: 4.1245e-06 Epoch 9/50 349/349 [==============================] - 3s 8ms/step - loss: 4.3127e-06 - val_loss: 4.0519e-06 Epoch 10/50 349/349 [==============================] - 3s 8ms/step - loss: 4.3005e-06 - val_loss: 4.5148e-06 Epoch 11/50 349/349 [==============================] - 3s 8ms/step - loss: 4.2540e-06 - val_loss: 3.9034e-06 Epoch 12/50 349/349 [==============================] - 3s 8ms/step - loss: 4.2584e-06 - val_loss: 3.7366e-06 Epoch 13/50 349/349 [==============================] - 3s 8ms/step - loss: 4.3550e-06 - val_loss: 3.9741e-06 Epoch 14/50 349/349 [==============================] - 3s 8ms/step - loss: 5.3136e-06 - val_loss: 1.1133e-05 Epoch 15/50 349/349 [==============================] - 3s 8ms/step - loss: 5.3412e-06 - val_loss: 4.8264e-06
Last modified: 17 November 2023