Comments (4)
Hey,
I was getting the same problem, and changing this line in upsample_to_size function:
y = im_size / x.shape[2]
to
y = im_size // x.shape[2]
helped me to solve this problem (as with the former / division, the result was a float rather than an integer).
from stylegan2-tensorflow-2.0.
TypeError Traceback (most recent call last)
<ipython-input-35-1135957cf48f> in <module>()
1 if __name__ == "__main__":
----> 2 model = StyleGAN(lr = 0.0001, silent = False)
3 model.evaluate(0)
4
5 while model.GAN.steps < 1000001:
14 frames
<ipython-input-34-e6decac87252> in __init__(self, steps, lr, decay, silent)
342
343 #Init GAN and Eval Models
--> 344 self.GAN = GAN(steps = steps, lr = lr, decay = decay)
345 self.GAN.GenModel()
346 self.GAN.GenModelA()
<ipython-input-34-e6decac87252> in __init__(self, steps, lr, decay)
161 #Init Models
162 self.discriminator()
--> 163 self.generator()
164
165 self.GMO = Adam(lr = self.LR, beta_1 = 0, beta_2 = 0.999)
<ipython-input-34-e6decac87252> in generator(self)
240 x = Reshape([4, 4, 4*cha])(x)
241
--> 242 x, r = g_block(x, inp_style[0], inp_noise, 32 * cha, u = False) #4
243 outs.append(r)
244
<ipython-input-34-e6decac87252> in g_block(inp, istyle, inoise, fil, u)
104 out = LeakyReLU(0.2)(out)
105
--> 106 return out, to_rgb(out, rgb_style)
107
108 def d_block(inp, fil, p = True):
<ipython-input-34-e6decac87252> in to_rgb(inp, style)
125 size = inp.shape[2]
126 x = Conv2DMod(3, 1, kernel_initializer = VarianceScaling(200/size), demod = False)([inp, style])
--> 127 return Lambda(upsample_to_size, output_shape=[None, im_size, im_size, None])(x)
128
129 def from_rgb(inp, conc = None):
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/base_layer.py in __call__(self, *args, **kwargs)
920 not base_layer_utils.is_in_eager_or_tf_function()):
921 with auto_control_deps.AutomaticControlDependencies() as acd:
--> 922 outputs = call_fn(cast_inputs, *args, **kwargs)
923 # Wrap Tensors in `outputs` in `tf.identity` to avoid
924 # circular dependencies.
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/layers/core.py in call(self, inputs, mask, training)
886 with backprop.GradientTape(watch_accessed_variables=True) as tape,\
887 variable_scope.variable_creator_scope(_variable_creator):
--> 888 result = self.function(inputs, **kwargs)
889 self._check_variables(created_variables, tape.watched_variables())
890 return result
<ipython-input-34-e6decac87252> in upsample_to_size(x)
75 def upsample_to_size(x):
76 y = im_size / x.shape[2]
---> 77 x = K.resize_images(x, y, y, "channels_last",interpolation='bilinear')
78 return x
79
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/backend.py in resize_images(x, height_factor, width_factor, data_format, interpolation)
2829 else:
2830 output_shape = (None, new_height, new_width, None)
-> 2831 x.set_shape(output_shape)
2832 return x
2833
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/ops.py in set_shape(self, shape)
637 # We want set_shape to be reflected in the C API graph for when we run it.
638 if not isinstance(shape, tensor_shape.TensorShape):
--> 639 shape = tensor_shape.TensorShape(shape)
640 dim_list = []
641 if shape.dims is None:
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/tensor_shape.py in __init__(self, dims)
769 self._dims = [as_dimension(dims)]
770 else:
--> 771 self._dims = [as_dimension(d) for d in dims_iter]
772
773 @property
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/tensor_shape.py in <listcomp>(.0)
769 self._dims = [as_dimension(dims)]
770 else:
--> 771 self._dims = [as_dimension(d) for d in dims_iter]
772
773 @property
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/tensor_shape.py in as_dimension(value)
714 return value
715 else:
--> 716 return Dimension(value)
717
718
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/tensor_shape.py in __init__(self, value)
198 TypeError("Dimension value must be integer or None or have "
199 "an __index__ method, got {!r}".format(value)),
--> 200 None)
201 if self._value < 0:
202 raise ValueError("Dimension %d must be >= 0" % self._value)
/usr/local/lib/python3.6/dist-packages/six.py in raise_from(value, from_value)
TypeError: Dimension value must be integer or None or have an __index__ method, got 256.0
from stylegan2-tensorflow-2.0.
Hey,
I was getting the same problem, and changing this line in upsample_to_size function:
y = im_size / x.shape[2]
to
y = im_size // x.shape[2]
helped me to solve this problem (as with the former / division, the result was a float rather than an integer).
Thank you so much for the reply. This worked!
I stumbled running model.load() though with the same problem appearing when it's loading the json. I tried changing some dtypes from the gen.json to int32, but it does not work, as well as turning all float32 to int32.
TypeError Traceback (most recent call last)
<ipython-input-26-31b4c636623b> in <module>()
6 # model.train()
7
----> 8 model.load(28)
9 n1 = noiseList(64)
10 n2 = nImage(64)
17 frames
<ipython-input-20-0bebf3a8d64a> in load(self, num)
614 self.GAN.D = self.loadModel("dis", num)
615 self.GAN.S = self.loadModel("sty", num)
--> 616 self.GAN.G = self.loadModel("gen", num)
617
618 self.GAN.GE = self.loadModel("genMA", num)
<ipython-input-20-0bebf3a8d64a> in loadModel(self, name, num)
595 file.close()
596
--> 597 mod = model_from_json(json, custom_objects = {'Conv2DMod': Conv2DMod})
598 mod.load_weights("/content/drive/My Drive/GAN/Models/"+name+"_"+str(num)+".h5")
599
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/saving/model_config.py in model_from_json(json_string, custom_objects)
114 config = json.loads(json_string)
115 from tensorflow.python.keras.layers import deserialize # pylint: disable=g-import-not-at-top
--> 116 return deserialize(config, custom_objects=custom_objects)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/layers/serialization.py in deserialize(config, custom_objects)
107 module_objects=globs,
108 custom_objects=custom_objects,
--> 109 printable_module_name='layer')
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/utils/generic_utils.py in deserialize_keras_object(identifier, module_objects, custom_objects, printable_module_name)
371 custom_objects=dict(
372 list(_GLOBAL_CUSTOM_OBJECTS.items()) +
--> 373 list(custom_objects.items())))
374 with CustomObjectScope(custom_objects):
375 return cls.from_config(cls_config)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/network.py in from_config(cls, config, custom_objects)
985 """
986 input_tensors, output_tensors, created_layers = reconstruct_from_config(
--> 987 config, custom_objects)
988 model = cls(inputs=input_tensors, outputs=output_tensors,
989 name=config.get('name'))
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/network.py in reconstruct_from_config(config, custom_objects, created_layers)
2027 if layer in unprocessed_nodes:
2028 for node_data in unprocessed_nodes.pop(layer):
-> 2029 process_node(layer, node_data)
2030
2031 input_tensors = []
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/network.py in process_node(layer, node_data)
1975 if input_tensors is not None:
1976 input_tensors = base_layer_utils.unnest_if_single_tensor(input_tensors)
-> 1977 output_tensors = layer(input_tensors, **kwargs)
1978
1979 # Update node index map.
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/base_layer.py in __call__(self, *args, **kwargs)
920 not base_layer_utils.is_in_eager_or_tf_function()):
921 with auto_control_deps.AutomaticControlDependencies() as acd:
--> 922 outputs = call_fn(cast_inputs, *args, **kwargs)
923 # Wrap Tensors in `outputs` in `tf.identity` to avoid
924 # circular dependencies.
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/layers/core.py in call(self, inputs, mask, training)
886 with backprop.GradientTape(watch_accessed_variables=True) as tape,\
887 variable_scope.variable_creator_scope(_variable_creator):
--> 888 result = self.function(inputs, **kwargs)
889 self._check_variables(created_variables, tape.watched_variables())
890 return result
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/layers/core.py in upsample_to_size(x)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/backend.py in resize_images(x, height_factor, width_factor, data_format, interpolation)
2829 else:
2830 output_shape = (None, new_height, new_width, None)
-> 2831 x.set_shape(output_shape)
2832 return x
2833
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/ops.py in set_shape(self, shape)
637 # We want set_shape to be reflected in the C API graph for when we run it.
638 if not isinstance(shape, tensor_shape.TensorShape):
--> 639 shape = tensor_shape.TensorShape(shape)
640 dim_list = []
641 if shape.dims is None:
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/tensor_shape.py in __init__(self, dims)
769 self._dims = [as_dimension(dims)]
770 else:
--> 771 self._dims = [as_dimension(d) for d in dims_iter]
772
773 @property
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/tensor_shape.py in <listcomp>(.0)
769 self._dims = [as_dimension(dims)]
770 else:
--> 771 self._dims = [as_dimension(d) for d in dims_iter]
772
773 @property
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/tensor_shape.py in as_dimension(value)
714 return value
715 else:
--> 716 return Dimension(value)
717
718
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/tensor_shape.py in __init__(self, value)
198 TypeError("Dimension value must be integer or None or have "
199 "an __index__ method, got {!r}".format(value)),
--> 200 None)
201 if self._value < 0:
202 raise ValueError("Dimension %d must be >= 0" % self._value)
/usr/local/lib/python3.6/dist-packages/six.py in raise_from(value, from_value)
TypeError: Dimension value must be integer or None or have an __index__ method, got 256.0
from stylegan2-tensorflow-2.0.
@cuakevinlex Hi. I ran into the same problem. I believe there is some compatibility issue with newer Tensorflows. I tested it using Tensorflow 2.1 and 2.2, both giving the same problem.
The author stated that he used Tensorflow 2.0 so I tried that out. It gave a different error regarding CPU/GPU. Turns out, TF 2.0 uses a different CUDA version. So I downloaded this Docker Image which should have compatible CUDA and TF versions. It ran perfectly.
from stylegan2-tensorflow-2.0.
Related Issues (20)
- the problem about official version HOT 2
- To better understanding
- Choice of loss function
- conv2d_mod/Conv2D NCHW not implemented HOT 2
- Finding latent vector for image
- Why two gradient tapes are used? HOT 1
- ValueError: Layer model_3 expects 8 input(s), but it received 9 input tensors. HOT 5
- Possible to generate 360 landscapes in equirectangular mode from exisiting NON-360 landcapes? HOT 1
- Can you make this work with Tensorflow 2.4.1 and above with tf.distribute.MirroredStrategy() HOT 2
- How to output single large images? HOT 1
- Asking the dataset and labels
- About model conversion HOT 2
- bad marshal data (unknown type code)
- Generated images are very white
- how to train a face editor with stylegan? HOT 2
- I keep hitting NaN value error. HOT 1
- load LandscapesBig mode error HOT 4
- Why in my training, not the same as Latent have the same generator pictures HOT 1
- loss problem HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from stylegan2-tensorflow-2.0.