brandontrabucco / design-bench

Benchmarks for Model-Based Optimization
MIT License
80 stars 19 forks source link

confused about which env to use : mbo or design-baselines? #17

Closed SourinDeyUW closed 7 months ago

SourinDeyUW commented 7 months ago

At the beginning of readme the instructions are given for installing mbo but later instructions are for design-benchlines. After I just install mbo, I could not run the Reproducing Baseline Performance codes. I could run (facing issues though) the codes after I install design-benchlines . My question is why should I install mbo if I am not using it at all.

brandontrabucco commented 7 months ago

Hi SourinDeyUW,

Thanks for your interest in our code!

design-baselines is a separate package that contains reference implementations of baselines. This needs to be installed if you want to use our baseline algorithms for benchmarking purposes.

git clone https://github.com/brandontrabucco/design-baselines
conda env create -f design-baselines/environment.yml
conda activate design-baselines

If you need these, then the mbo environment is not required. Otherwise, if you don't need to use the baselines, then you only need to create the base mbo environment (which has fewer packages installed than design-baselines).

Also, mbo is just a conda environment name we chose for illustration purposes, feel free to choose your own!