https://github.com/atharva-naik/sync

Implementation for SYNC: Structurally guided Hard Negative Curricula for Generalizable Neural Code Search
https://github.com/atharva-naik/sync

Last synced: 30 days ago
JSON representation

Implementation for SYNC: Structurally guided Hard Negative Curricula for Generalizable Neural Code Search

• Host: GitHub
• URL: https://github.com/atharva-naik/sync
• Owner: atharva-naik
• License: mit
• Created: 2023-09-17T00:58:15.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2024-03-16T23:42:57.000Z (about 1 year ago)
• Last Synced: 2025-02-05T04:08:30.377Z (3 months ago)
• Language: Python
• Size: 6.32 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# **SYNC**

Implementation for SYNC: Structurally guided Hard Negative Curricula for Generalizable Neural Code Search

## Environment

To recreate our computational environment we provide a YML file export of our conda environment (py3.7.yml)

```

conda env create -f py3.7.yml

```

## **Data:**

You can download all the files from here: https://drive.google.com/drive/folders/1LFm1lrjKxYHAm9Dq7fKx8m0ZNtRv88wI?usp=sharing 

(put in data folder).

and 

`triple` files from here: https://drive.google.com/drive/folders/18796d0_F9hpv6sV7-2RBQA4DMMVxLlVL?usp=sharing (put in triples folder).

Download the pre-generated negative samples for SYNC () and DISCO ().

## **Running Instructions:**

Running instructions for training models and baselines. 


(**NOTE:** for the ```_ood.sh``` scripts just adjust the experiment folder path or ```-en``` argument)

### **Baselines:**

Simple baselines:

**nBOW:**

```

bash scripts/train_nbow.sh

bash scripts/nbow_ood.sh

```

**RNN:**

```

bash scripts/train_rnn.sh

bash scripts/rnn_ood.sh

```

**CNN:**

```

bash scripts/train_cnn.sh

bash scripts/cnn_ood.sh

```

**Regular triplet training:**

CodeBERT:

```

bash scripts/codebert.sh

bash scripts/codebert_ood.sh

```

GraphCodeBERT:

```

bash scripts/graphcodebert.sh

bash scripts/graphcodebert_ood.sh

```

UniXcoder:

```

bash scripts/unixcoder.sh

bash scripts/unixcoder_ood.sh

```

**DNS:**

CodeBERT:

```

bash scripts/codebert_dns.sh

bash scripts/codebert_ood.sh

```

GraphCodeBERT:

```

bash scripts/graphcodebert_dns.sh

bash scripts/graphcodebert_ood.sh

```

UniXcoder:

```

bash scripts/unixcoder_dns.sh

bash scripts/unixcoder_ood.sh

```

**DISCO:**

CodeBERT:

```

bash scripts/codebert_disco.sh

bash scripts/codebert_ood.sh

```

GraphCodeBERT:

```

bash scripts/graphcodebert_disco.sh

bash scripts/graphcodebert_ood.sh

```

UniXcoder:

```

bash scripts/unixcoder_disco.sh

bash scripts/unixcoder_ood.sh

```

CodeRetriever:

CodeBERT:

```

bash scripts/codebert_code_retriever.sh

bash scripts/codebert_ood.sh

```

GraphCodeBERT:

```

bash scripts/graphcodebert_code_retriever.sh

bash scripts/graphcodebert_ood.sh

```

UniXcoder:

```

bash scripts/unixcoder_code_retriever.sh

bash scripts/unixcoder_ood.sh

```

### **SYNC experiments:**

CodeBERT:

```

bash scripts/codebert_sync.sh

bash scripts/codebert_ood.sh

```

GraphCodeBERT:

```

bash scripts/graphcodebert_sync.sh

bash scripts/graphcodebert_ood.sh

```

UniXcoder:

```

bash scripts/unixcoder_sync.sh

bash scripts/unixcoder_ood.sh

```

## **Citation:**

```

@inproceedings{naik23sync,

    title = {SYNC: A Structurally guided Hard Negative Curricula for

Generalizable Neural Code Search},

    author = {Atharva Naik and Soumtira Das and Jyothi Vedurada and Somak Aditya},

    booktitle = {The Asia-Pacific Chapter of the Association for Computational Linguistics},

    month = {November},

    year = {2023}

}

```