ICSME2022CrossCS

Code and data for ICSME 2022 research track paper "Cross-Modal Contrastive Learning for Code Search"

Dataset

You can obtain the dataset directly from this link. Note that you may need to "gzip -d" the data file to "*.jsonl".

Dependency

• python>=3.6
• torch==1.4.0
• transformers>=2.5.0

Fine-tune

sh run.sh

Inference

sh test.sh

Evaluator

python evaluator/evaluator.py -a dataset/test.jsonl  -p saved_models/predictions.jsonl