search

HDI-Project / ATM

Auto Tune Models - A multi-tenant, multi-data system for automated machine learning (model selection and tuning).
https://hdi-project.github.io/ATM/
MIT License
525 stars 141 forks source link

Adapt tuner instantiation for BTB keyword processing #122

Closed micahjsmith closed 5 years ago

micahjsmith commented 5 years ago

Quickstart fails with this error (reported by @csala):

  File "/home/xals/Projects/Pythia/MIT/ATM/atm/worker.py", line 379, in run_classifier
    params = self.tune_hyperparameters(hyperpartition)
  File "/home/xals/Projects/Pythia/MIT/ATM/atm/worker.py", line 189, in tune_hyperparameters
    r_minimum=self.datarun.r_minimum)
TypeError: __init__() got an unexpected keyword argument 'r_minimum'

The worker's self.Tuner is an instance of btb.tuning.base.BaseTuner, but the different subclasses take different arguments.

Implement a atm.utilities.make_tuner adapter that passes only the relevant kwargs.

See also:

beevabeeva commented 5 years ago

Hi there. I'm posting here in case its the same issue: After executing python scripts/worker.py I get the following output, repeatedly till killing the process :

Computing on datarun 1
Selector: <class 'btb.selection.uniform.Uniform'>
Tuner: <class 'btb.tuning.uniform.Uniform'>
Error choosing hyperparameters: datarun=<ID = 1, dataset ID = 1, strategy = uniform__uniform, budget = classifier (100), status: running>
Traceback (most recent call last):
  File "/home/research/ATM2/atm/atm-env/lib/python2.7/site-packages/atm-0.1.1.dev0-py2.7.egg/atm/worker.py", line 379, in run_classifier
    params = self.tune_hyperparameters(hyperpartition)
  File "/home/research/ATM2/atm/atm-env/lib/python2.7/site-packages/atm-0.1.1.dev0-py2.7.egg/atm/worker.py", line 189, in tune_hyperparameters
    r_minimum=self.datarun.r_minimum)
TypeError: __init__() got an unexpected keyword argument 'r_minimum'

Something went wrong. Sleeping 1 seconds.

I tried changing to the stable git branch, but I get the same result.

Thanks. Edit: I just read the rest of the docs page and I'm pretty sure this is expected behaviour. I will let it run for a while. Sorry.

csala commented 5 years ago

Hi there. I'm posting here in case its the same issue:

It is, indeed, the same error.

This should be fixed once the PR #124 is merged.

Regarding stable branch, make sure to have switched to it before installing everything! If you checkout master and install from there and then switch to stable you will still have the dependencies from master installed and it will fail anyway.

If you checkout to stable and then create a fresh virtualenv and install, this error should go away.

beevabeeva commented 5 years ago

Thanks so much, again @csala !