【AutoTVM】How to use graph_opt.log on other model after auto-tuning

I use auto-tuning , get the xx_graph_opt.log.

When I relay.build_module.build other model, but use the default log(~/.tvm/tophub/llvm_v0.03.log).

I want to use the xx_graph_opt.log instead of the default log,how do I do?

Each auto-tuning will take dozens of hours,I don’t want to auto-tune every time for different models.