我的模型在训练时给出了这个答案。准确度仅以二进制形式出现 - 0 或 1。此外，我的模型仅预测 D 标签（[0 0 0 1]）。
string[] lineParts = line.Split(new[] { ' ' }, 2);
string timestamp = lineParts[0];
string phrase = lineParts[1];
这是我的模型：
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses, lstm_cell_2_layer_call_fn while saving (showing 5 of 15). These functions will not be directly callable after loading.
INFO:tensorflow:Assets written to: /content/drive/MyDrive/SavedModels/OpenfaceRNN_final_7/assets
INFO:tensorflow:Assets written to: /content/drive/MyDrive/SavedModels/OpenfaceRNN_final_7/assets
[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
[[0. 0. 1. 0.]]
(1, 2362, 715)
Epoch 1/50
1/1 [==============================] - 1s 729ms/step - loss: 1.4395 - accuracy: 0.0000e+00
Epoch 2/50
1/1 [==============================] - 1s 696ms/step - loss: 1.4322 - accuracy: 0.0000e+00
Epoch 3/50
1/1 [==============================] - 1s 705ms/step - loss: 1.4250 - accuracy: 0.0000e+00
Epoch 4/50
1/1 [==============================] - 1s 707ms/step - loss: 1.4178 - accuracy: 0.0000e+00
Epoch 5/50
1/1 [==============================] - 1s 699ms/step - loss: 1.4107 - accuracy: 0.0000e+00
Epoch 6/50
1/1 [==============================] - 1s 703ms/step - loss: 1.4036 - accuracy: 0.0000e+00
Epoch 7/50
1/1 [==============================] - 1s 708ms/step - loss: 1.3965 - accuracy: 0.0000e+00
Epoch 8/50
1/1 [==============================] - 1s 712ms/step - loss: 1.3895 - accuracy: 0.0000e+00
Epoch 9/50
1/1 [==============================] - 1s 700ms/step - loss: 1.3826 - accuracy: 0.0000e+00
Epoch 10/50
1/1 [==============================] - 1s 709ms/step - loss: 1.3756 - accuracy: 0.0000e+00
Epoch 11/50
1/1 [==============================] - 1s 697ms/step - loss: 1.3688 - accuracy: 0.0000e+00
Epoch 12/50
1/1 [==============================] - 1s 698ms/step - loss: 1.3619 - accuracy: 0.0000e+00
Epoch 13/50
1/1 [==============================] - 1s 796ms/step - loss: 1.3551 - accuracy: 0.0000e+00
Epoch 14/50
1/1 [==============================] - 1s 783ms/step - loss: 1.3483 - accuracy: 0.0000e+00
Epoch 15/50
1/1 [==============================] - 1s 727ms/step - loss: 1.3416 - accuracy: 0.0000e+00
Epoch 16/50
1/1 [==============================] - 1s 705ms/step - loss: 1.3349 - accuracy: 0.0000e+00
Epoch 17/50
1/1 [==============================] - 1s 702ms/step - loss: 1.3282 - accuracy: 0.0000e+00
Epoch 18/50
1/1 [==============================] - 1s 709ms/step - loss: 1.3215 - accuracy: 0.0000e+00
Epoch 19/50
1/1 [==============================] - 1s 695ms/step - loss: 1.3149 - accuracy: 0.0000e+00
Epoch 20/50
1/1 [==============================] - 1s 700ms/step - loss: 1.3083 - accuracy: 0.0000e+00
Epoch 21/50
1/1 [==============================] - 1s 706ms/step - loss: 1.3017 - accuracy: 0.0000e+00
Epoch 22/50
1/1 [==============================] - 1s 703ms/step - loss: 1.2952 - accuracy: 0.0000e+00
Epoch 23/50
1/1 [==============================] - 1s 701ms/step - loss: 1.2887 - accuracy: 0.0000e+00
Epoch 24/50
1/1 [==============================] - 1s 717ms/step - loss: 1.2822 - accuracy: 0.0000e+00
Epoch 25/50
1/1 [==============================] - 1s 709ms/step - loss: 1.2757 - accuracy: 0.0000e+00
Epoch 26/50
1/1 [==============================] - 1s 702ms/step - loss: 1.2692 - accuracy: 0.0000e+00
Epoch 27/50
1/1 [==============================] - 1s 707ms/step - loss: 1.2628 - accuracy: 0.0000e+00
Epoch 28/50
1/1 [==============================] - 1s 732ms/step - loss: 1.2564 - accuracy: 0.0000e+00
Epoch 29/50
1/1 [==============================] - 1s 705ms/step - loss: 1.2500 - accuracy: 0.0000e+00
Epoch 30/50
1/1 [==============================] - 1s 702ms/step - loss: 1.2436 - accuracy: 0.0000e+00
Epoch 31/50
1/1 [==============================] - 1s 707ms/step - loss: 1.2373 - accuracy: 0.0000e+00
Epoch 32/50
1/1 [==============================] - 1s 708ms/step - loss: 1.2309 - accuracy: 0.0000e+00
Epoch 33/50
1/1 [==============================] - 1s 696ms/step - loss: 1.2246 - accuracy: 0.0000e+00
Epoch 34/50
1/1 [==============================] - 1s 710ms/step - loss: 1.2183 - accuracy: 0.0000e+00
Epoch 35/50
1/1 [==============================] - 1s 703ms/step - loss: 1.2120 - accuracy: 0.0000e+00
Epoch 36/50
1/1 [==============================] - 1s 715ms/step - loss: 1.2058 - accuracy: 1.0000
Epoch 37/50
1/1 [==============================] - 1s 711ms/step - loss: 1.1995 - accuracy: 1.0000
Epoch 38/50
1/1 [==============================] - 1s 709ms/step - loss: 1.1933 - accuracy: 1.0000
Epoch 39/50
1/1 [==============================] - 1s 701ms/step - loss: 1.1870 - accuracy: 1.0000
Epoch 40/50
1/1 [==============================] - 1s 706ms/step - loss: 1.1808 - accuracy: 1.0000
Epoch 41/50
1/1 [==============================] - 1s 725ms/step - loss: 1.1746 - accuracy: 1.0000
Epoch 42/50
1/1 [==============================] - 1s 710ms/step - loss: 1.1685 - accuracy: 1.0000
Epoch 43/50
1/1 [==============================] - 1s 697ms/step - loss: 1.1623 - accuracy: 1.0000
Epoch 44/50
1/1 [==============================] - 1s 697ms/step - loss: 1.1561 - accuracy: 1.0000
Epoch 45/50
1/1 [==============================] - 1s 712ms/step - loss: 1.1500 - accuracy: 1.0000
Epoch 46/50
1/1 [==============================] - 1s 701ms/step - loss: 1.1439 - accuracy: 1.0000
Epoch 47/50
1/1 [==============================] - 1s 710ms/step - loss: 1.1378 - accuracy: 1.0000
Epoch 48/50
1/1 [==============================] - 1s 711ms/step - loss: 1.1317 - accuracy: 1.0000
Epoch 49/50
1/1 [==============================] - 1s 699ms/step - loss: 1.1256 - accuracy: 1.0000
Epoch 50/50
1/1 [==============================] - 1s 704ms/step - loss: 1.1195 - accuracy: 1.0000