在 Keras 中使用自定义精度度量时出错

Question

from keras.models import Sequential
from keras.layers import Dense, Dropout
import matplotlib.pyplot as plt
import seaborn as sns
from matplotlib.pyplot import figure

model = Sequential()
model.add(Dense(90, input_dim=900, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(90, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(10, activation='relu'))
model.add(Dense(1, activation='sigmoid'))

m = keras.metrics.Precision(class_id=1)
# example data of suitable dimension, to offer MRE to SO
X_train = np.eye(900)
Y_train = np.ones((900, 1))

model.compile(loss='binary_crossentropy', optimizer='adam', metrics=[m])
model.fit(X_train, Y_train, epochs=50, batch_size=1000)

When I use model.compile(loss='binary_crossentropy, optimizer='adam', metrics=['Precision'])

it works, but when I use Precision(class_id=1) (regardless of whether I substitute it as a variable), I get

ValueError: slice index 1 of dimension 1 out of bounds. 
for '{{node strided_slice_1}} = StridedSlice[Index=DT_INT32, T=DT_FLOAT, begin_mask=0, ellipsis_mask=1, end_mask=0, 
new_axis_mask=0, shrink_axis_mask=2](Cast_1, strided_slice_1/stack, strided_slice_1/stack_1, strided_slice_1/stack_2)' 
with input shapes: [?,1], [2], [2], [2] and with computed 
input tensors: input[1] = <0 1>, input[2] = <0 2>, input[3] = <1 1>.

I don't know what any of this stuff means. Slice of WHAT is out of bounds? (I defined X_train and Y_train, of course, and they work when I just write metric=['Precision']). FYI I'm doing this in SageMaker, if it makes any difference. There is no other code, so if I am failing to define some config thing, I don't know about that.

Answer 1

As far as I can tell (im not confident), your data doesn't seem to be long enough? Like i think its having an issue trying to slice your data because it cant find an index 1 in the first dimension, meaning there is 1 or 0 items of data.

Try checking your data and see if what you expect is there.