TypeError: ('Bad input argument to theano function with name "holographic_memory.py:84"

[SeekPoint]

envy@ub1404:~/os_pri/github/Associative_LSTM$ python holographic_memory.py Using gpu device 0: GeForce GTX 950M (CNMeM is disabled, CuDNN 4007) Traceback (most recent call last): File "holographic_memory.py", line 94, in MEM = write_func(KEYS, VALUES) File "/home/envy/.local/lib/python2.7/site-packages/theano/compile/function_module.py", line 786, in call allow_downcast=s.allow_downcast) File "/home/envy/.local/lib/python2.7/site-packages/theano/tensor/type.py", line 139, in filter raise TypeError(err_msg, data) TypeError: ('Bad input argument to theano function with name "holographic_memory.py:84" at index 0(0-based)', 'TensorType(float32, matrix) cannot store a value of dtype float64 without risking loss of precision. If you do not mind this loss, you can: 1) explicitly cast your data to float32, or 2) set "allow_input_downcast=True" when calling "function".', array([[-0.33998727, 0.77906697, 0.06991387, ..., -0.98946589, -0.83468061, 0.89153633], [-0.63970228, 0.8195924 , 0.9473632 , ..., 0.16236229, 0.94564315, 0.97183497], [ 0.67610139, 0.99756132, 0.8097881 , ..., -0.69396546, 0.8336102 , -0.98980493], ..., [ 0.27303741, 0.97723419, -0.99634828, ..., -0.42349166, 0.35726541, 0.99777589], [ 0.56724025, 0.79452556, 0.90919262, ..., -0.50166936, -0.97753518, -0.96124023], [ 0.28472163, 0.10105307, -0.1343447 , ..., -0.51666857, 0.95856671, -0.89232788]])) envy@ub1404:~/os_pri/github/Associative_LSTM$

[mpezeshki]

Using THEANO_FLAGS='floatX=float32' python holographic_memory.py should fix the problem.

[SeekPoint]

I tried:

envy@ub1404:~/os_pri/github/Associative_LSTM$ THEANO_FLAGS='floatX=float32' python holographic_memory.py Using gpu device 0: GeForce GTX 950M (CNMeM is disabled, CuDNN 4007) Traceback (most recent call last): File "holographic_memory.py", line 94, in MEM = write_func(KEYS, VALUES) File "/home/envy/.local/lib/python2.7/site-packages/theano/compile/function_module.py", line 786, in call allow_downcast=s.allow_downcast) File "/home/envy/.local/lib/python2.7/site-packages/theano/tensor/type.py", line 139, in filter raise TypeError(err_msg, data) TypeError: ('Bad input argument to theano function with name "holographic_memory.py:84" at index 0(0-based)', 'TensorType(float32, matrix) cannot store a value of dtype float64 without risking loss of precision. If you do not mind this loss, you can: 1) explicitly cast your data to float32, or 2) set "allow_input_downcast=True" when calling "function".', array([[ 0.85184374, 0.82000731, 0.99867008, ..., 0.43216262, -0.28698177, -0.26221031], [ 0.8731134 , -0.26070533, -0.91038367, ..., -0.06170114, -0.19895874, -0.13945744], [ 0.3685904 , -0.99917839, -0.90257495, ..., 0.47001036, 0.66174286, 0.12531638], ..., [-0.81111444, 0.94144517, -0.90505509, ..., 0.73322565, 0.8069545 , 0.94837876], [ 0.46073917, 0.40346909, -0.98421055, ..., -0.70648842, 0.2854518 , 0.53658445], [ 0.9140164 , 0.82759999, -0.52525026, ..., 0.36035819, 0.03019263, 0.82799066]])) envy@ub1404:~/os_pri/github/Associative_LSTM$

and

envy@ub1404:~/os_pri/github/Associative_LSTM$ THEANO_FLAGS='allow_input_downcast=True' python holographic_memory.py /home/envy/.local/lib/python2.7/site-packages/theano/configdefaults.py:1686: UserWarning: Theano does not recognise this flag: allow_input_downcast warnings.warn('Theano does not recognise this flag: {0}'.format(key)) Using gpu device 0: GeForce GTX 950M (CNMeM is disabled, CuDNN 4007) Traceback (most recent call last): File "holographic_memory.py", line 94, in MEM = write_func(KEYS, VALUES) File "/home/envy/.local/lib/python2.7/site-packages/theano/compile/function_module.py", line 786, in call allow_downcast=s.allow_downcast) File "/home/envy/.local/lib/python2.7/site-packages/theano/tensor/type.py", line 139, in filter raise TypeError(err_msg, data) TypeError: ('Bad input argument to theano function with name "holographic_memory.py:84" at index 0(0-based)', 'TensorType(float32, matrix) cannot store a value of dtype float64 without risking loss of precision. If you do not mind this loss, you can: 1) explicitly cast your data to float32, or 2) set "allow_input_downcast=True" when calling "function".', array([[ 0.55820402, -0.99933052, -0.58131817, ..., -0.99162348, 0.97391971, 0.99976266], [-0.64940681, -0.06587449, -0.49446634, ..., 0.76723741, -0.99872607, 0.57148139], [ 0.98469854, 0.98922329, 0.68558021, ..., -0.54050539, -0.97274425, 0.13653712], ..., [-0.86314311, -0.78690848, -0.95348783, ..., 0.87118976, 0.84511147, -0.47139609], [ 0.99752764, -0.66803953, 0.84460468, ..., -0.97220982, 0.45180846, 0.22152956], [-0.54236742, 0.05867093, -0.97811148, ..., 0.88242457, 0.93022328, -0.98206891]])) envy@ub1404:~/os_pri/github/Associative_LSTM$

[SeekPoint]

it doesn't work yet