I was looking at optuna package and noticed that they raise this ValueError:if step is not None: if log: raise ValueError("The parameter
stepis not supported when
logis True.")
So step_size
should always be None
Hi LethalDolphin75
I think you are right there isn't one (although I remember a discussion about it...)
Anyhow it will be very easy to implement, just inherit from:
https://github.com/allegroai/clearml/blob/400c6ec103d9f2193694c54d7491bb1a74bbe8e8/clearml/automation/parameters.py#L111
And return the power of the parent value here:
https://github.com/allegroai/clearml/blob/400c6ec103d9f2193694c54d7491bb1a74bbe8e8/clearml/automation/parameters.py#L146
And
https://github.com/allegroai/clearml/blob/400c6ec103d9f2193694c54d7491bb1a74bbe8e8/clearml/automation/parameters.py#L158
Wdyt?
Hi AgitatedDove14 , I made the PR: https://github.com/allegroai/clearml/pull/462 . Check it out and let me know. Thanks!
I'm having another problem now because I am using the OptunaOptimizer.
Hmm let me check a sec
LethalDolphin75 Yes you are correct, we should add here:
https://github.com/allegroai/clearml/blob/400c6ec103d9f2193694c54d7491bb1a74bbe8e8/clearml/automation/optuna/optuna.py#L210elif isinstance(p, UniformLogarithmicParameterRange): hp_type = 'suggest_float' hp_params = dict(low=p.min_value, high=p.max_value if p.include_max else p.max_value - p.step_size, log=True, step=p.step_size)
btw: I'm not sure if the min/max values should be before or after log (i.e. low=log(p.min_value) or low=p.min_value )
Hey LethalDolphin75 , when it works, could you PR it?
Looks great, nice solution and easy to implement, thanks!
And the low and high values are already the log value. For example it couldbe sth like low=1e-6, high=1e-3
; and not low=-6, high=-3
.
I'm having another problem now because I am using the OptunaOptimizer.