Requirements

• torch==1.7.1
• numpy==1.23.3
• transformers==4.23.1 (You can also use the static version in this repo)
• sentencepiece==0.1.97
• tokenizer==0.13.1
• sacremoses==0.0.53
• tqdm==4.64.1

Quick Start

• Download the pre-trained multilingual language model mBERT or XLM-R
• To quickly reproduce the results with French (fr) as the target langauge and mBERT (mbert) as the backbone under the supervised setting:

python main.py --tfm_type mbert --tgt_lang fr

sh run_qpen.sh

• To reproduce other results, ref to the data_utils.py for details

Usage

To run experiments under different settings, change the exp_type setting:

• supervised refers to the supervised setting
• acs is the proposed method

Two example scripts:

• run_qpen.sh provides an example to run basic experiment.