如何从张量获得预测输出

Question

Currently, I've finished my functional neuron model from TensorFlow, and I want to output a prediction to the CSV file.

However, when I get the prediction, it's a tensor with 8 classes output probability, 428 rows.

        0           1           2           3           4          5            6          7
0   0.001258    0.971390    0.000170    0.005542    0.000077    0.002042    0.001015    0.018505
1   0.000663    0.989145    0.000010    0.000393    0.000642    0.000798    0.003435    0.004914
2   0.001139    0.939257    0.000232    0.017595    0.000112    0.000762    0.000535    0.040368
3   0.002857    0.970207    0.000041    0.000680    0.003005    0.000273    0.012651    0.010286
4   0.000706    0.992306    0.000017    0.000533    0.000908    0.000095    0.000808    0.004627
.
.
428

How can I grab the highest probability of each class number as output? For example, the first data point 0, it's clear to see the class 1 has the highest propablility.

Answer 1

Use argmax() method to get the index with the largest value across axes of a tensor.

b = [0, 1, 2, 100, 3, 4, 5]
print(np.argmax(b))

3

Tensorflow Argmax Doc

Numpy Argmax Doc