Instruction for executing Python code with a specific GPU and save console outputs and error to a log file

run_gpu_python.sh

# Use `nvtop` or `nvidia-smi` to check avalible GPUs. Only allows the Python code to access a specific GPU:
# CUDA_VISIBLE_DEVICES=<GPU_INDEX>

# Redirect standard output to `main.log` and then redirect standard error to standard output:
# > main.log 2>&1

# Allows code to run in the bakcgounrd:
# &
CUDA_VISIBLE_DEVICES=0 python main.py > main.log 2>&1 &

# Check CPU usage:
top

# Check GPU usage:
nvtop

# Kill program:
kill <UID>

# To read the end of log file
tail -f main.log