Issue on manipulate_latent(manipulate_model, (x_test, y_test), args) at tf 2.2 gpu

[dolannachina]

I train capsule at --epochs 50 --batch_size 400 --routings 3 --lam_recon 0.392
then got a "trained_model.h5"

but when python capsulenet.py -t -w result/trained_model.h5
I will got message:

Model: "functional_1"

---

Layer (type) Output Shape Param # Connected to

input_1 (InputLayer) [(400, 28, 28, 1)] 0

---

conv1 (Conv2D) (400, 20, 20, 256) 20992 input_1[0][0]

---

primarycap_conv2d (Conv2D) (400, 6, 6, 256) 5308672 conv1[0][0]

---

primarycap_reshape (Reshape) (400, 1152, 8) 0 primarycap_conv2d[0][0]

---

primarycap_squash (Lambda) (400, 1152, 8) 0 primarycap_reshape[0][0]

---

digitcaps (CapsuleLayer) (400, 10, 16) 1474560 primarycap_squash[0][0]

---

input_2 (InputLayer) [(None, 10)] 0

---

mask (Mask) (400, 160) 0 digitcaps[0][0]
input_2[0][0]

---

capsnet (Length) (400, 10) 0 digitcaps[0][0]

---

decoder (Sequential) (None, 28, 28, 1) 1411344 mask[0][0]

Total params: 8,215,568
Trainable params: 8,215,568
Non-trainable params: 0

---

end!
------------------------------Begin: manipulate------------------------------
2020-12-20 05:58:34.387299: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library cublas64_10.dll
2020-12-20 05:58:34.628204: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library cudnn64_7.dll
2020-12-20 05:58:35.453403: W tensorflow/stream_executor/gpu/redzone_allocator.cc:314] Internal: Invoking GPU asm compilation is supported on Cuda non-Windows platforms only
Relying on driver to perform ptx compilation.
Modify $PATH to customize ptxas location.
This message will be only logged once.
Traceback (most recent call last):
File "C:/CapsNet-Keras-tf2.2/capsulenet.py", line 273, in
manipulate_latent(manipulate_model, (x_test, y_test), args)
File "C:/python/CapsNet-Keras-tf2.2/capsulenet.py", line 184, in manipulate_latent
x_recon = model.predict([x, y, tmp])
File "N:\anaconda3\lib\site-packages\tensorflow\python\keras\engine\training.py", line 130, in _method_wrapper
return method(self, *args, **kwargs)
File "N:\anaconda3\lib\site-packages\tensorflow\python\keras\engine\training.py", line 1599, in predict
tmp_batch_outputs = predict_function(iterator)
File "N:\anaconda3\lib\site-packages\tensorflow\python\eager\def_function.py", line 780, in call
result = self._call(*args, **kwds)
File "N:\anaconda3\lib\site-packages\tensorflow\python\eager\def_function.py", line 846, in _call
return self._concrete_stateful_fn._filtered_call(canon_args, canon_kwds) # pylint: disable=protected-access
File "N:\anaconda3\lib\site-packages\tensorflow\python\eager\function.py", line 1848, in _filtered_call
cancellation_manager=cancellation_manager)
File "N:\anaconda3\lib\site-packages\tensorflow\python\eager\function.py", line 1924, in _call_flat
ctx, args, cancellation_manager=cancellation_manager))
File "N:\anaconda3\lib\site-packages\tensorflow\python\eager\function.py", line 550, in call
ctx=ctx)
File "N:\anaconda3\lib\site-packages\tensorflow\python\eager\execute.py", line 60, in quick_execute
inputs, attrs, num_outputs)
tensorflow.python.framework.errors_impl.InvalidArgumentError: Input to reshape is a tensor with 9216 values, but the requested shape requires a multiple of 3200
[[node functional_5/primarycap_reshape/Reshape (defined at C:/CapsNet-Keras-tf2.2/capsulenet.py:184) ]] [Op:__inference_predict_function_992]

Function call stack:
predict_function

How to fix this issue?

[ghost]

I had the same error.
How about making the batch_size on line 180 of capsule.py explicit, like in the test function?

x_recon = model.predict([x, y, tmp], batch_size=1)