Open sandptel opened 1 year ago
Try Improving the model's accuracy by testing different configurations for the layers and note down the accuracies for them. Conv1 (Conv2D) (None, 50, 198, 32) 320 ['image[0][0]']
pool1 (MaxPooling2D) (None, 25, 99, 32) 0 ['Conv1[0][0]']
Conv2 (Conv2D) (None, 25, 99, 64) 18496 ['pool1[0][0]']
pool2 (MaxPooling2D) (None, 12, 49, 64) 0 ['Conv2[0][0]']
reshape (Reshape) (None, 12, 3136) 0 ['pool2[0][0]']
dense1 (Dense) (None, 12, 64) 200768 ['reshape[0][0]']
dropout_1 (Dropout) (None, 12, 64) 0 ['dense1[0][0]']
bidirectional_2 (Bidirectional (None, 12, 256) 197632 ['dropout_1[0][0]'] )
bidirectional_3 (Bidirectional (None, 12, 128) 164352 ['bidirectional_2[0][0]'] )
Change these parameter test out different results.
Test out the accuracies by testing on the kaggle notebook provided.
Try Improving the model's accuracy by testing different configurations for the layers and note down the accuracies for them. Conv1 (Conv2D) (None, 50, 198, 32) 320 ['image[0][0]']
pool1 (MaxPooling2D) (None, 25, 99, 32) 0 ['Conv1[0][0]']
Conv2 (Conv2D) (None, 25, 99, 64) 18496 ['pool1[0][0]']
pool2 (MaxPooling2D) (None, 12, 49, 64) 0 ['Conv2[0][0]']
reshape (Reshape) (None, 12, 3136) 0 ['pool2[0][0]']
dense1 (Dense) (None, 12, 64) 200768 ['reshape[0][0]']
dropout_1 (Dropout) (None, 12, 64) 0 ['dense1[0][0]']
bidirectional_2 (Bidirectional (None, 12, 256) 197632 ['dropout_1[0][0]']
)
bidirectional_3 (Bidirectional (None, 12, 128) 164352 ['bidirectional_2[0][0]']
)
Change these parameter test out different results.