I’ve a numpy array that I’m trying to load into a pytorch tensor, it’s loaded properly on cpu
device but on cuda
device it fails with following error;
>> torch.tensor(X_train, dtype=torch.float64, device=cuda)
RuntimeError Traceback (most recent call last)
<ipython-input-92-4ee38d4004b1> in <module>()
----> 1 torch.tensor(X_train, dtype=torch.float64, device=cuda)
RuntimeError: cuda runtime error (59) : device-side assert triggered at /pytorch/aten/src/THC/generic/THCTensorCopy.cpp:20
I tried to load the numpy array into a pandas DataFrame to have a look at it but nothing seems suspecious:
>> pd.DataFrame(X_train).describe()
0 1 2 3 4 5 6 7 8 9 ... 560 561 562 563 564 565 566 567 568 569
count 30.000000 30.000000 30.000000 30.000000 30.000000 30.000000 30.000000 30.000000 30.000000 30.000000 ... 30.000000 30.000000 30.000000 30.000000 30.000000 30.000000 30.000000 30.000000 30.000000 30.000000
mean -0.611895 -0.610648 -0.606697 -0.601020 -0.593792 -0.585476 -0.577100 -0.568880 -0.561298 -0.555327 ... -0.976487 -0.977011 -0.977404 -0.977851 -0.978500 -0.979447 -0.979884 -0.979993 -0.980466 -0.980602
std 0.007103 0.007094 0.007025 0.006931 0.007014 0.007098 0.007005 0.006994 0.007035 0.007000 ... 0.006211 0.006145 0.006196 0.006212 0.006270 0.006313 0.006392 0.006372 0.006364 0.006407
min -0.622784 -0.622222 -0.619049 -0.613251 -0.606484 -0.598921 -0.590348 -0.582025 -0.574668 -0.568545 ... -0.995849 -0.996575 -0.997365 -0.997696 -0.998277 -0.999678 -1.000187 -1.000031 -1.000642 -1.001061
25% -0.615420 -0.614054 -0.609959 -0.604032 -0.596716 -0.588481 -0.579605 -0.571327 -0.563762 -0.557526 ... -0.978997 -0.979154 -0.979423 -0.979826 -0.980450 -0.981289 -0.982201 -0.982305 -0.982654 -0.982617
50% -0.611913 -0.610543 -0.606757 -0.601161 -0.593762 -0.585272 -0.577268 -0.569677 -0.561556 -0.555934 ... -0.975869 -0.976176 -0.976744 -0.977298 -0.977871 -0.978881 -0.979175 -0.979325 -0.979627 -0.980093
75% -0.609884 -0.608681 -0.604591 -0.599247 -0.591120 -0.582454 -0.574215 -0.565808 -0.558239 -0.551438 ... -0.972170 -0.973330 -0.973860 -0.974376 -0.974811 -0.975666 -0.976206 -0.976166 -0.976381 -0.976528
max -0.590921 -0.589520 -0.586062 -0.582049 -0.575003 -0.566571 -0.559088 -0.551611 -0.544998 -0.539241 ... -0.968274 -0.968620 -0.968871 -0.969230 -0.970103 -0.971065 -0.971451 -0.971528 -0.971549 -0.971597
8 rows × 570 columns
What could be the reason of this error?