Reward Imputation with Sketching for Contextual Batched Bandits

This repository is the official implementation of Reward Imputation with Sketching for Contextual Batched Bandits.

This paper was accepted by NeurIPS 2023.

Requirements

To install requirements:

pip install -r requirements.txt

Training

To train models except DFM-S in the paper, run this command:

python algo_main.py  Algorithm_name

To train DFM-S in the paper, run this command:

python algo_main2.py

We recommend you tuning hyper-parameters by using nni module. In our experiments, we use nni to tune hyper-parameters.

Evaluation

Because we calculate average reward in each episode, you can export reward data using nni after running code. To export reward data, run:

nnictl experiment export [experiment_id] --filename [file_path] --type json --intermediate

Connect

If you have any questions, please contact us at the email address [email protected], or submit an issue here.