I want to train my dataset with YOLO in Jupyter Notebook and I need to use the GPU. I am encountering errors while training the model.

code:

model.train(data=data_config_path, epochs=5, optimizer='Adam', batch=8, imgsz=640,lr0=0.01, device=0)

error:

ValueError: Invalid CUDA 'device=0' requested. Use 'device=cpu' or pass valid CUDA device(s) if available, i.e. 'device=0' or 'device=0,1,2,3' for Multi-GPU.

torch.cuda.is_available(): False
torch.cuda.device_count(): 0
os.environ['CUDA_VISIBLE_DEVICES']: None
Seehttps://pytorch.org/get-started/locally/ for up-to-date torch install instructions if no CUDA devices are seen by torch.

To solve the problem:

I downloaded and installed the CUDA Toolkit 11.8 Downloads* exe file.

• From the PyTorch website**, I ran the installation command: pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118.

• From the CUDNN web site***, I downloaded the cuDNN, Windows zip file and copied the bin, include, and lib folders from the zip file to the respective folders in “C:Program FilesNVIDIA GPU Computing ToolkitCUDAv11.8”.

Additionally, I added the following paths to the environment variables:

• C:Program FilesNVIDIA GPU Computing ToolkitCUDAv11.8bin

• C:Program FilesNVIDIA GPU Computing ToolkitCUDAv11.8libnvvp

Now, when I check the version information in the terminal using nvcc --version, it shows the correct version information.

import torch
print(torch.cuda.is_available())
print(torch.cuda.device_count())
print(torch.version.cuda)

When I run the following commands in PowerShell:

• I get the results True, 1, 11.8 respectively.

However, when I run these same commands in Jupyter Notebook,

I get False, 0, None.

I couldn’t figure out why I get True in PowerShell but False in Jupyter Notebook.

*(https://developer.nvidia.com/cuda-11-8-0-download-archive)

**(https://pytorch.org/get-started/locally/ )

***(https://developer.nvidia.com/rdp/cudnn-archive)

****”Download cuDNN v8.9.7 (December 5th, 2023), for CUDA 11.x”

I want to train my dataset with YOLO in Jupyter Notebook