Model Time Reduction using torch 2.0 compile

1

I have been using hidet backend for compiling pytorch model. Iam using the torch 2.0 for reducing the appgen time for a model run.

Specify better ways to reduce model run time by using torch.compile.
hidet.torch.dynamo_config.search_space(2)
with torch.no_grad():
hidet.torch.dynamo_config.print_input_graph(True)
hidet.torch.dynamo_config.correctness_report()
pytorch_model = torch.compile(pytorch_model,backend = ‘hidet’)

    return pytorch_model