Install¶

PyPI¶

pip install pymltoolkit

If the installation failed with dependancy issues, execute the above command with –no-dependencies

pip install pymltoolkit --no-dependencies

Setup TensorFlow with GPU support (Optional)¶

Refer the official TensorFlow documentation (https://www.tensorflow.org/install/gpu) for most up to date innstructions.

PyMLToolKit is tested with the following software versions in Windows 10

CUDA Toolkit 10.0 (10.0.130_411.31_win10)
cuDNN SDK (v7.4.2.24)

Step #1

Install latest NVIDIA® GPU drivers
Install CUDA Toolkit
Install cuDNN SDK
Set System Path to CUDA Toolkit. If the CUDA Toolkit is installed to “C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.0” and extracted cuDNN content to r”C:/Program FilesNVIDIA GPU Computing Toolkit/cuDNN”, update your %PATH% to match:

SET PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.0\bin;%PATH%
SET PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.0\extras\CUPTI\libx64;%PATH%
SET PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.0\include;%PATH%
SET PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\cuDNN\bin;%PATH%

Step #2

Install TensorFLow-GPU

PyPI

pip install tensorflow-gpu

To install specific version

pip install tensorflow-gpu==1.14

Check GPU in Tensorflow (output forat as below)

from tensorflow.python.client import device_lib
print(device_lib.list_local_devices())

[name: "/device:CPU:0"
device_type: "CPU"
memory_limit: 99999999
locality {
}
incarnation: 9999999999, name: "/device:GPU:0"
device_type: "GPU"
memory_limit: 99999999
locality {
   bus_id: 1
   links {
   }
}
incarnation: 99999999
physical_device_desc: "device: 0, name: XXXXXX, pci bus id: 0000:00:00.0, compute capability: 0.0"]

memory_limit is in bytes. To convert allocated memeory to GB use : memory_limit/(1024*1024*1024)

If you encounter errors in setting up TensorFlow, please refer to thw official TensorFlow Build and install error messages (https://www.tensorflow.org/install/errors)