README.md

metadata

library_name: transformers
pipeline_tag: text-classification
tags:
  - text-classification
  - pytorch
  - jax
  - code_x_glue_cc_defect_detection
  - code
  - roberta
  - security
  - vulnerability-detection
  - codebert
  - apache-2.0
license: apache-2.0

CodeBERT fine-tuned for Java Vulnerability Detection

CodeBERT model fine-tuned for detecting security vulnerabilities in Java code.

Model Description

This model is fine-tuned from microsoft/codebert-base for binary classification of secure/insecure Java code.

Intended Uses

• Detect security vulnerabilities in Java source code
• Binary classification: Safe (LABEL_0) vs Vulnerable (LABEL_1)

How to Use

from transformers import AutoTokenizer, AutoModelForSequenceClassification

tokenizer = AutoTokenizer.from_pretrained("mangsense/codebert_java")
model = AutoModelForSequenceClassification.from_pretrained("mangsense/codebert_java")

# run code
```python
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
import numpy as np
tokenizer = AutoTokenizer.from_pretrained('mrm8488/codebert-base-finetuned-detect-insecure-code')
model = AutoModelForSequenceClassification.from_pretrained('mrm8488/codebert-base-finetuned-detect-insecure-code')

inputs = tokenizer("your code here", return_tensors="pt", truncation=True, padding='max_length')
labels = torch.tensor([1]).unsqueeze(0)  # Batch size 1
outputs = model(**inputs, labels=labels)
loss = outputs.loss
logits = outputs.logits

print(np.argmax(logits.detach().numpy()))

Training Data

Trained on CodeXGLUE Defect Detection dataset.

Limitations

• Focused on Java code only
• May not detect all types of vulnerabilities