chejuju/multi-task_priority

PRIMA

source code for the paper : A Tale of Two Tasks: Automated Issue Priority Prediction with Deep Multi-task Learning .

Requirement

1. python 3.7
2. pytorch 1.1

How to run?

1. You should download "pytorch_model.bin" file ( a pre-trained Bert model ) from https://huggingface.co/lanwuwei/BERTOverflow_stackoverflow_github/tree/main, and put it into the " bert_pretrain " folder of this project

2. Run "python run.py --model muti_bert_rcnn_att" to train Prima (default: kubernetes dataset)

3. if you want to change dataset, modify the first code line of run.py.

4. if you want to change model and mode, you will need to change run.py and the commend.

For example,

change "from muti_train_eval import train" to "from priority_train_eval import train" in run.py ,

meanwhile, Run "python run.py --model priority_bert_rcnn_att" to train single priority task of PRIMA.