Closed jaehoonkoo closed 2 years ago
Hi,
Thank you for asking. The MLA_Clean function was previously used to differentiate between MLA that does not load historical data and MLA that loads historical data. The MLA_Clean function is now removed. Instead, users can invoke MLA with no loading by using meta data options (e.g. "load_func_eval": "no"). The issue was because the GPTune-Demo TLA example code was still using the old interface.
I have updated the GPTune-Demo TLA example code and the TLA experiment script. This will use the new interface. https://github.com/gptune/GPTune/commit/c56aeab92d28d7e4d6b81836aa8d94761fcd5463 https://github.com/gptune/GPTune/commit/870406bf098cbf1076de3a6843cc8777d56f4c6c
Could you checkout the latest version and try the example again? Thank you!
Hi @jke513,
In the commit https://github.com/gptune/GPTune/commit/c56aeab92d28d7e4d6b81836aa8d94761fcd5463, I accidentally disabled the default db data load option.
So, to use a more stable version of GPTune, please also apply this commit https://github.com/gptune/GPTune/commit/b3075688dca2c67bd2e7d4f264d67b9989da6f37. Running a single "git pull" command will be enough. Thank you!
Thank you for the edit.
Hi,
I am trying to run TLA code. https://github.com/gptune/GPTune/blob/master/examples/GPTune-Demo/demo_TLA.py#L255
AttributeError: 'GPTune' object has no attribute 'MLA_Clean'
Is MLA_Clean() available in GPTune??Thank you for your help.