vinx13 opened a new pull request #5724: URL: https://github.com/apache/incubator-tvm/pull/5724
We need to clear compiler engine cache after task extraction, otherwise after tuning, we have the compilation result for task extraction in the cache and we won't utilize the latest configs from after tuning. This can be reproduced by running any AutoTVM tutorial such as https://github.com/apache/incubator-tvm/blob/master/tutorials/autotvm/tune_relay_cuda.py cc @comaniac @merrymercy ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org