How to ensure that auto-generated Triton Kernel is executed

torch.compile
1

Hi all, is there a way to ensure that the auto-generated Triton kernels are executed at runtime. I have set the following environment variables. Which log file to look at in order to ensure that if the auto-generated Triton code is executed versus native PyTorch kernel code. Thanks.

export TORCH_LOGS=“output_code”
export TORCH_COMPILE_DEBUG=1
export TORCH_COMPILE_VERBOSE=1
export INDUCTOR_LOG_LEVEL=“INFO”