ObjectiveONNX: Objective interface for ONNX Models.

In slds-lmu/paper_2021_multi_fidelity_surrogates: Surrogate benchmarks for HPO problems

Description

Objective interface where user can pass an R function that works on an 'data.table()'.

Super classes

bbotk::Objective -> bbotk::ObjectiveRFunDt -> ObjectiveONNX

Methods

Public methods

• ObjectiveONNX$new()

• ObjectiveONNX$clone()

Inherited methods

• bbotk::Objective$eval()

• bbotk::Objective$format()

• bbotk::Objective$print()

• bbotk::ObjectiveRFunDt$eval_dt()

• bbotk::ObjectiveRFunDt$eval_many()

---

Method new()

Creates a new instance of this [R6][R6::R6Class] class.

Usage

ObjectiveONNX$new(
  model_path,
  data_order,
  trafo_dict,
  domain,
  full_codomain,
  codomain,
  task = NULL,
  id = "ONNX",
  active_session = TRUE,
  retrafo = FALSE,
  properties = character(),
  constants = NULL,
  check_values = FALSE
)

Arguments

model_path

('character')
Path to the onnx Model.

data_order

('character')
Order of columns in data.

trafo_dict

('list')
Dictionary containing feature transformations before beeing fed to the NN.

full_codomain

([paradox::ParamSet])
Full codomain.

task

('character(1)')
Name of a task to be fixed.

id

('character(1)').

active_session

('logical(1)')
Whether an active session should be used.

retrafo

('logical(1)')
Should params be trafoed back to their original range before return?

properties

('character()').

---

Method clone()

The objects of this class are cloneable with this method.

Usage

ObjectiveONNX$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

slds-lmu/paper_2021_multi_fidelity_surrogates documentation built on Feb. 20, 2022, 11:53 a.m.