CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutSign UpSign In
amanchadha

Real-time collaboration for Jupyter Notebooks, Linux Terminals, LaTeX, VS Code, R IDE, and more,
all in one place. Commercial Alternative to JupyterHub.

GitHub Repository: amanchadha/coursera-deep-learning-specialization
 Path: blob/master/C5 - Sequence Models/Week 1/Jazz improvisation with LSTM/inference_code.py
Views: 4819
1
def inference_model(LSTM_cell, densor, n_x = 90, n_a = 64, Ty = 100):

2
    """

3
    Uses the trained "LSTM_cell" and "densor" from model() to generate a sequence of values.

4
    

5
    Arguments:

6
    LSTM_cell -- the trained "LSTM_cell" from model(), Keras layer object

7
    densor -- the trained "densor" from model(), Keras layer object

8
    n_x -- number of unique values

9
    n_a -- number of units in the LSTM_cell

10
    Ty -- number of time steps to generate

11
    

12
    Returns:

13
    inference_model -- Keras model instance

14
    """

15
    

16
    # Define the input of your model with a shape 

17
    x0 = Input(shape=(1, n_x))

18
    

19
    # Define s0, initial hidden state for the decoder LSTM

20
    a0 = Input(shape=(n_a,), name='a0')

21
    c0 = Input(shape=(n_a,), name='c0')

22
    a = a0

23
    c = c0

24
    x = x0

25


26
    ### START CODE HERE ###

27
    # Step 1: Create an empty list of "outputs" to later store your predicted values (≈1 line)

28
    outputs = []

29
    

30
    # Step 2: Loop over Ty and generate a value at every time step

31
    for t in range(Ty):

32
        

33
        # Step 2.A: Perform one step of LSTM_cell (≈1 line)

34
        a, _, c = LSTM_cell(x, initial_state=[a, c])

35
        

36
        # Step 2.B: Apply Dense layer to the hidden state output of the LSTM_cell (≈1 line)

37
        out = densor(a)

38


39
        # Step 2.C: Append the prediction "out" to "outputs" (≈1 line)

40
        outputs.append(out)

41
        

42
        # Step 2.D: Set the prediction "out" to be the next input "x". You will need to use RepeatVector(1). (≈1 line)

43
        x = RepeatVector(1)(out)

44
        

45
    # Step 3: Create model instance with the correct "inputs" and "outputs" (≈1 line)

46
    inference_model = Model(inputs=[x0, a0, c0], outputs=outputs)

47
    ### END CODE HERE ###

48
    

49
    return inference_model

50


51


52
inference_model = inference_model(LSTM_cell, densor)

53


54


55
x1 = np.zeros((1, 1, 90))

56
x1[:,:,35] = 1

57
a1 = np.zeros((1, n_a))

58
c1 = np.zeros((1, n_a))

59
predicting = inference_model.predict([x1, a1, c1])

60


61


62
indices = np.argmax(predicting, axis = -1)

63
results = to_categorical(indices, num_classes=90)

64