adhwiraj/security-scanner-llm

Security Scanner LLM (PEFT LoRA Adapter)

This is a PEFT LoRA adapter for the microsoft/phi-2 model, fine-tuned for security vulnerability detection.

Model Details

• Base Model: microsoft/phi-2
• PEFT Type: LoRA
• Task: Causal Language Modeling
• Training Dataset: adhwiraj/security-scanner-dataset-private

Usage

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base_model_name = "microsoft/phi-2"
adapter_name = "adhwiraj/security-scanner-llm"

# Load base model
model = AutoModelForCausalLM.from_pretrained(base_model_name)
tokenizer = AutoTokenizer.from_pretrained(base_model_name)

# Load PEFT adapter
model = PeftModel.from_pretrained(model, adapter_name)

# Use for inference
prompt = "Analyze this code for security vulnerabilities:\n\n```python\ndef query(user_input):\n    return f'SELECT * FROM users WHERE name = {user_input}'\n```\n\nSecurity analysis:"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=200)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)

Training

This model was trained using PEFT LoRA (Low-Rank Adaptation) technique with the following parameters:

• r: 8
• lora_alpha: 16
• lora_dropout: 0.1
• target_modules: ["fc1", "out_proj", "fc2", "Wqkv"]