Skip to content

Latest commit

 

History

History
30 lines (15 loc) · 1.15 KB

File metadata and controls

30 lines (15 loc) · 1.15 KB

Informed Meta-Learning with INPs

i also try here to use the INP framework for a biologicla problem (predicting ligand protein binding)

Setup

The environment.yaml lists the required packages to reproduce the experiments presented in the paper. To install the evnironment run:

conda env create -f environment.yaml

conda activate inps

Experiments with Synthetic Data

jobs/run_sinusoids.sh contais commands that need to be run to reproduce the experiments with synthetic data

After training the models, results can be analyzed with the following two notebooks:

  • evaluation/evaluate_sinusoids.ipynb contains the analysis of the base experiments
  • evaluation/evaluate_sinusoids_dist_shift.ipynb contains the analysis of the train/test distribution shift experiment

Experiments with the Tempereatures Dataset

jobs/run_temperatures.sh contais commands that need to be run to reproduce the experiments with the tempereatures datasets

After training the models, results can be analyzes with evaluation/evaluate_temperature.ipynb