Code for reproducing Learning-To-Measure: In-Context Active Feature Acquisition
We recommend using Conda with the provided environment.yml file.
conda env create -f environment.ymlBegin by pretraining predictor on synthetic tasks:
python main.py --experiment sim --mode trainOnce the predictor is trained, pretrain the policy on synthetic tasks:
python main.py --experiment sim --mode train_afaEvaluate acquisition performance:
python main.py --experiment sim --mode test_afa