theanorc.txt

[global]
floatX = float32
cast_policy = numpy+floatX
device = cuda
force_device = True
optimizer_including = cudnn : local_remove_all_assert
\\cxx = C:\Users\ninja\anaconda3\pkgs\m2w64-gcc-5.3.0-6\Library\mingw-w64\bin\g++.exe
optimizer_excluding = more_mem
openmp = False

[cuda]
root = C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.4

[dnn]
enabled = auto
library_path = C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.4\lib\x64
include_path = C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.4\include

[blas]
ldflags = -lmkl_rt.1 -lcublas64_70 

[magma]
enabled = True
library_path = C:\Users\ninja\anaconda3\pkgs\magma-2.5.4-h154cb12_2\Library\lib
include_path = C:\Users\ninja\anaconda3\pkgs\magma-2.5.4-h154cb12_2\Library\include

[gpuarray]
preallocate = 0.75
sched = multi

[scan]
allow_gc = True
allow_output_prealloc = True