https://github.com/sumny/iaml

Reproduce iaml results
https://github.com/sumny/iaml

Last synced: 18 days ago
JSON representation

Reproduce iaml results

• Host: GitHub
• URL: https://github.com/sumny/iaml
• Owner: sumny
• License: gpl-3.0
• Created: 2022-05-27T08:57:01.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2022-06-03T19:56:50.000Z (over 2 years ago)
• Last Synced: 2024-12-17T13:46:41.340Z (23 days ago)
• Language: R
• Size: 33.2 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# iaml

Reproduce results from YAHPO Gym iaml.

## Installation

You can install the released version from GitHub using:

``` r

remotes::install_github("sumny/iaml")

```

Note that for *exact* reproduction, you need the R packages with the exact version as specified in the DESCRIPTION

(i.e., == instead of >=) and also R version 4.0.5.

Also note that some measures (ram and time related ones) can never be exactly reproduced because they depend on system

hardware and system load.

By default, in all `eval_*` functions, the original fixed seed is used and therefore, runs are fully deterministic,

(except for the ram and time measures mentioned above).

To change this, simply provide a custom seed.

Note that data is obtained via `mlr3oml` and is cached via `qs`.

## Example

```r

library(iaml)

scenario = "iaml_glmnet"

configuration = list(alpha = 0.1, s = 0.1, trainsize = 0.05, task_id = "40981")

eval_yahpo(scenario, configuration)

```