Neuraxle
Neuraxle copied to clipboard
The world's cleanest AutoML library ✨ - Do hyperparameter tuning with the right pipeline abstractions to write clean deep learning production pipelines. Let your pipeline steps have hyperparameter sp...
General improvements and bugfixes: - Update FlattenForEach that wasn't up to date - Use _ids, data_inputs, and expected_outputs more often in DataContainer instead of shorthand ids, di, and eo, which...
**Is your feature request related to a problem? Please describe.** How to design complex execution flows around pipelines? **Describe the solution you'd like** Creating the following steps in Neuraxle to...
We need `def HyperparamSamples.get_nullified_keys` to remember which hyperparams were nullified by the Optional and ChooseOneOf classes, ChoosManyOf, etc. ``` hp = step.get_hyperparams() dict_with_double_underscores = hp.to_flat_dict() the_keys_of_the_double_underscores_that_were_nullified_this_time = hp.get_nullified_keys() dict_with_double_underscores.remove(the_keys_of_the_double_underscores_that_were_nullified_this_time) ```...
Sometimes the usage of a hyperparameters defined for a step within the pipeline depends on another hyperparameter. We should have a way to know if which hyperparameters are used by...
Feature: TrialMetricsPlottingObserver should support various plot functions for various metric type
**Is your feature request related to a problem? Please describe.** TrialMetricsPlottingObserver is a neat subscriber-listener interface designed to be used with a BaseHyperparamsRepository instance. However, it assumes that every metric...
The problem: Currently the neuraxle BaseStep has a fit method signature with only 2 parameters (data_inputs, expected_outputs). In libraries like [keras](https://github.com/keras-team/keras/blob/master/keras/engine/training.py#L1215) it is possible to have additional arguments being passed...
I think the Pipeline class has been written before ForceHandleMixin was a thing. I believe inheritance of the class could make it way more simple than it is now.
**Is your feature request related to a problem? Please describe.** I think that there is either an error with Neuraxle's LogUniform, or else with [scipy's LogUniform](https://github.com/scipy/scipy/pull/10815 ). **Describe the solution...
I think it would be a neat addition to generate Discrete-value hyperparams. You could redefine Boolean (which should really be called Bernouilli in my opinion) to be Binomial(1,p). In the...
**Is your feature request related to a problem? Please describe.** Adding spaces would return wrappers of the two spaces. If encountering a float or an int, it'd wrap it by...