I tried giving the input as input channel and output channel as tf.constant tensor. But the code shows error as input for the class must be a vector of {int32,int64}, got shape [2,1]. I defined the GTLayer class as follows :
Importing libraries:
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
import numpy as np
GTLayer class:
class GTLayer(keras.layers.Layer):
def __init__(self, in_channels, out_channels, first=True):
super(GTLayer, self).__init__()
self.in_channels = in_channels
self.out_channels = out_channels
self.first = first
if self.first == True:
self.conv1 = GTConv(in_channels, out_channels)
self.conv2 = GTConv(in_channels, out_channels)
else:
self.conv1 = GTConv(in_channels, out_channels)
def forward(self, A, H_=None):
if self.first == True:
a = self.conv1(A)
b = self.conv2(A)
#H = torch.bmm(a,b)
H = tf.matmul(a, b)
#W = [(F.softmax(self.conv1.weight, dim=1)).detach(),(F.softmax(self.conv2.weight, dim=1)).detach()]
W = [tf.stop_gradient(tf.nn.softmax(self.conv1.weight, axis=1).numpy()),
tf.stop_gradient(tf.nn.softmax(self.conv1.weight, axis=1).numpy()) ]
else:
a = self.conv1(A)
#H = torch.bmm(H_,a)
H = tf.matmul(H_, a)
#W = [(F.softmax(self.conv1.weight, dim=1)).detach()]
W = [tf.stop_gradient(tf.nn.softmax(self.conv1.weight, axis=1).numpy())]
return H,W
GTConv layer
class GTConv(keras.layers.Layer):
def __init__(self, in_channels, out_channels):
super(GTConv, self).__init__()
self.in_channels = in_channels
self.out_channels = out_channels
w_init = tf.random_normal_initializer()
self.weight = tf.Variable(
initial_value=w_init(shape=(in_channels, out_channels)),
trainable=True)
self.bias = None
self.scale = tf.Variable([0.1] , trainable=False)
self.reset_parameters()
#self.weight = nn.Parameter(torch.Tensor(out_channels,in_channels,1,1))
#self.bias = None
#self.scale = nn.Parameter(torch.Tensor([0.1]), requires_grad=False)
def reset_parameters(self):
n = self.in_channels
tf.fill(self.weight, 9)
#if self.bias is not None:
# fan_in, _ = nn.init._calculate_fan_in_and_fan_out(self.weight)
# bound = 1 / math.sqrt(fan_in)
# nn.init.uniform_(self.bias, -bound, bound)
def forward(self, A):
A = tf.add_n(tf.nn.softmax(self.weight))
return A
Error:
input:
inp = tf.constant([4])
out = tf.constant([2])
d = GTLayer(inp, out)
--------------------------------------------------------------------------- InvalidArgumentError Traceback (most recent call last) in () ----> 1 d = GTLayer(inp, out)
5 frames /usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/ops.py in raise_from_not_ok_status(e, name) 7184 def raise_from_not_ok_status(e, name): 7185 e.message += (" name: " + name if name is not None else “”) → 7186 raise core._status_to_exception(e) from None # pylint: disable=protected-access 7187 7188
InvalidArgumentError: shape must be a vector of {int32,int64}, got shape [2,1] [Op:RandomStandardNormal]