we1k/Randeng-MLT-PromptCBLUE

CCKS2023-PromptCBLUE: Code implement of TianChi completition

This repo is the code implement of TianChi Competition: CCKS2023-PromptCBLUE (Ranked 3 / 381 )

Overview

On based of Chinese MLT Pretrained model Randeng-MLT, we finetune it on a Chinese Multitask medical Seq2Seq dataset, PromptCBLUE. Also added verbaliser for better and faster model convergence.

Getting Started:

1. Collect PromptCBLUE data under aug_data folder, then run

python data_utils/verbaliser.py

2. Start training!

bash scripts/train.sh

Acknowledgements:

Pretrained model: IDEA-CCNL/Randeng-T5-784M-MultiTask-Chinese [https://huggingface.co/IDEA-CCNL/Randeng-T5-784M-MultiTask-Chinese]