Examples: query, "exact match", wildcard*, wild?ard, wild*rd
Fuzzy search: cake~ (finds cakes, bake)
Term boost: "red velvet"^4, chocolate^2
Field grouping: tags:(+work -"fun-stuff")
Escaping: Escape characters +-&|!(){}[]^"~*?:\ with \, e.g. \+
Range search: properties.timestamp:[1587729413488 TO *] (inclusive), properties.title:{A TO Z}(excluding A and Z)
Combinations: chocolate AND vanilla, chocolate OR vanilla, (chocolate OR vanilla) NOT "vanilla pudding"
Field search: properties.title:"The Title" AND text
Answered
Hi. When Using Sklearn'S

Hi. When using sklearn's RandomizedSearchCV or GridSearchCV is there a way to save in a single clearml experiment the evaluation metrics of each hyperparameters set while the search is running?

  
  
Posted 2 years ago
Votes Newest

Answers 8


I am using the Task.init() approach
and then running RandomizedSearchCV
(not using ClearML's HPO).

Trying that now passing verbose=3
to sklearn's class.
I can see the verbose message on the clearml's console tab while the search runs so this is a kind of a poor's man solution to my problem buy may be enough for now.

Regarding using clearml HPO, will it create multiple experiments on the UI for each tested hyperparameters set or would i be able to see all those trials in a single experiment.
This is a curial point for using clearml HPO since comparing dozens of experiments in the UI and searching for the best is just not manageable.

  
  
Posted 2 years ago

DistressedGoat23 , how are you running this hyper parameter tuning? Ideally you need to have

` From clearml import Task

task = Task.init() `
In your running code, from that point onwards you should have tracking

  
  
Posted 2 years ago

Or are you using ClearML's HPO?

  
  
Posted 2 years ago

We are running a hyperparameter tuning (using some cv) which might take a long time and might be even aborted unexpectedly due to machine resources.
We therefore want to see the progress. what hyperparameters set were tested and what was their results summary metrics (i.e avg and stddev of ROCAUC across all cv's).

  
  
Posted 2 years ago

Hi DistressedGoat23 , can you please elaborate a bit on what you're like to do?

  
  
Posted 2 years ago

DistressedGoat23 check this example:
https://github.com/allegroai/clearml/blob/master/examples/optimization/hyper-parameter-optimization/hyper_parameter_optimizer.py
aSearchStrategy = RandomSearchIt will collect everything on the main Task

This is a curial point for using clearml HPO since comparing dozens of experiments in the UI and searching for the best is just not manageable.

You can of course do that (notice you can actually order them by scalars they report, and even do nested sorting, by holding the shift and selecting additional columns to sort by)
But it might be easier to collect on a single Task as in the example, as you pointed out 🙂
Any reason not to do that this way ?

  
  
Posted 2 years ago

Thanks! I will take a look and try.

  
  
Posted 2 years ago

DistressedGoat23

We are running a hyperparameter tuning (using some cv) which might take a long time and might be even aborted unexpectedly due to machine resources.
We therefore want to see the progress

On the HPO Task itself (not the individual experiments the one controlling it all) there is the global progress of the optimization metric, is this what you are looking for ? Am I missing something?

  
  
Posted 2 years ago
911 Views
8 Answers
2 years ago
one year ago
Tags