Build model...
Traceback (most recent call last):
File "ram.py", line 117, in
model = RecurrentAttentionMemory()
File "ram.py", line 86, in init
x = shared_attention((memory, x))
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/keras/_impl/keras/engine/base_layer.py", line 239, in call
output = super(Layer, self).call(inputs, *kwargs)
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/layers/base.py", line 714, in call
outputs = self.call(inputs, args, *kwargs)
File "/Users/tushar/workspace/python/impact/songyouwei/SA-DL/attention_layer.py", line 66, in call
score = K.cast(mask, K.floatx())
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/keras/_impl/keras/backend.py", line 1226, in cast
return math_ops.cast(x, dtype)
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/ops/math_ops.py", line 779, in cast
x = ops.convert_to_tensor(x, name="x")
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 950, in convert_to_tensor
as_ref=False)
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 1040, in internal_convert_to_tensor
ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref)
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/ops/array_ops.py", line 1026, in _autopacking_conversion_function
return _autopacking_helper(v, inferred_dtype, name or "packed")
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/ops/array_ops.py", line 988, in _autopacking_helper
constant_op.constant(elem, dtype=dtype, name=str(i)))
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/framework/constant_op.py", line 214, in constant
value, dtype=dtype, shape=shape, verify_shape=verify_shape))
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/framework/tensor_util.py", line 421, in make_tensor_proto
raise ValueError("None values not supported.")
ValueError: None values not supported.
Here is the traceback:
Build model... Traceback (most recent call last): File "ram.py", line 117, in
model = RecurrentAttentionMemory()
File "ram.py", line 86, in init
x = shared_attention((memory, x))
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/keras/_impl/keras/engine/base_layer.py", line 239, in call
output = super(Layer, self).call(inputs, *kwargs)
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/layers/base.py", line 714, in call
outputs = self.call(inputs, args, *kwargs)
File "/Users/tushar/workspace/python/impact/songyouwei/SA-DL/attention_layer.py", line 66, in call
score = K.cast(mask, K.floatx())
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/keras/_impl/keras/backend.py", line 1226, in cast
return math_ops.cast(x, dtype)
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/ops/math_ops.py", line 779, in cast
x = ops.convert_to_tensor(x, name="x")
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 950, in convert_to_tensor
as_ref=False)
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 1040, in internal_convert_to_tensor
ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref)
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/ops/array_ops.py", line 1026, in _autopacking_conversion_function
return _autopacking_helper(v, inferred_dtype, name or "packed")
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/ops/array_ops.py", line 988, in _autopacking_helper
constant_op.constant(elem, dtype=dtype, name=str(i)))
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/framework/constant_op.py", line 214, in constant
value, dtype=dtype, shape=shape, verify_shape=verify_shape))
File "/usr/local/lib/python3.6/site-packages/tensorflow/python/framework/tensor_util.py", line 421, in make_tensor_proto
raise ValueError("None values not supported.")
ValueError: None values not supported.
Is there something that needs fixing?