Code of the paper Learning to Tokenize for Generative Retrieval.
pytorch, transformers, accelerate, faiss, k_means_constrained
NQ320K: unzip dataset/nq320k.zip
Other datasets coming soon.
Code for GenRet on NQ320K:
python run.py --model_name t5-base --code_num 512 --max_length 3 --train_data dataset/nq320k/train.json --dev_data dataset/nq320k/dev.json --corpus_data dataset/nq320k/corpus_lite.json --save_path out/model
Code for generative retrieval baselines: baseline.py
Code for dense retrieval baselines: dpr.py
@article{Sun2023LearningTT,
title={Learning to Tokenize for Generative Retrieval},
author={Weiwei Sun and Lingyong Yan and Zheng Chen and Shuaiqiang Wang and Haichao Zhu and Pengjie Ren and Zhumin Chen and Dawei Yin and M. de Rijke and Zhaochun Ren},
journal={ArXiv},
year={2023},
volume={abs/2304.04171},
}