The official pytorch implementation for our paper UNBERT: User-News Matching BERT for News Recommendation
pip install -r requirements.txt
For the MIND dataset, please download at https://msnews.github.io
File Name | Description |
---|---|
pretrainedModel | pretrained model from huggingface, including config.json, pytorch_model.bin and vocab.txt |
small/train | MIND-small train dataset, including behaviors.tsv and news.tsv |
small/dev | MIND-small dev dataset |
large/train | MIND-large train dataset |
large/dev | MIND-large dev dataset |
large/test | MIND-large test dataset |
python run.py --mode train --split small --root ./data/ --pretrain data/bert-base-uncased/
For more parameter settings, please refer to run.py
.