The dataset(PLCVD) in dataset_combined.xlsx comprises 8,800 pairs of Chinese-Vietnamese place names and 726 pairs of Chinese-Vietnamese personal names. Other subset of PLCVD are collected in this directory.

chin.json the dictionray of Chinese character and Sino-Vietnamese.
train.json and valid.json Used for finetuning models.
xlsx2json.py .xlsx to .json

• Under final/test directory

    Terminolgy-level seen data --> valid250.json
    Terminolgy-level unseen data --> unseen.json
    Sentence-level data --> test1600_wobracket.json

Check our report for more detail.

Our experiment results are stored in this directory.

• gemini

  Results of Gemini

• mBART

  Results of mBART

• llama2

  Results of Llama 2 7B

• english_data_result

  Results of Llama 2 7B with english data

Scripts for data preprocessing, training, evaluation are put here.

• baseline

  Scripts for using Gemini API. For more detail, check this.

• eval

  Scripts for evaluation. For more detail, check this.

python mbart.py for using our mbart model
python llama22.py for using our llama 2 model