Skip to content

Issue with MaxPooling1D #969

Closed
Closed
@AdrienDeverin

Description

@AdrienDeverin

Dear Community,
I have a problem when trying to add a MaxPooling1D layer in any of my models.

Here an exemple which work in python but not in c# ():

var t = tf.constant(new int[] { 1, 2, 3, 4, 5 });
t = tf.reshape(t, (1, 5, 1));
Tensor max_pool_1d = keras.layers.MaxPooling1D(pool_size: 2, strides: 1, padding : "valid").Apply(t);
NDArray array = max_pool_1d.numpy();

I'm really confused by the error I get (l.3) :

Tensorflow.InvalidArgumentError : 'Length for attr 'strides' of 3 must be at least minimum 4
; NodeDef: {{node MaxPool}}; Op<name=MaxPool; signature=input:T -> output:T; attr=T:type,default=DT_FLOAT,allowed=[DT_HALF, DT_BFLOAT16, DT_FLOAT, DT_DOUBLE, DT_INT32, DT_INT64, DT_UINT8, DT_INT16, DT_INT8, DT_UINT16, DT_QINT8]; attr=ksize:list(int),min=4; attr=strides:list(int),min=4; attr=padding:string,allowed=["SAME", "VALID", "EXPLICIT"]; attr=explicit_paddings:list(int),default=[]; attr=data_format:string,default="NHWC",allowed=["NHWC", "NCHW", "NCHW_VECT_C"]>

Even if I didn't understand why the argument "strides" should be >= 4, change it to this value send the same mistake.

Note : I didn't get any problems in the case of 2D or even 3D pooling.

Thanks,

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions