cssupport/mobilebert-sql-injection-detect

Model Card for Model ID

Based on google/mobilebert-uncased (MobileBERT is a thin version of BERT_LARGE, while equipped with bottleneck structures and a carefully designed balance between self-attentions and feed-forward networks). This model detects SQLInjection attacks in the input string (check How To Below). This is a very very light model (100mb) and can be used for edge computing use cases. Used dataset from Kaggle called SQl_Injection. Please test the model before deploying into any environment. Contact us for more info: [email protected]

Code Repo

Here is the code repo https://github.com/cssupport23/AI-Model---SQL-Injection-Attack-Detector

Model Details

Model Description

• Developed by: cssupport ([email protected])
• Model type: Language model
• Language(s) (NLP): English
• License: Apache 2.0
• Finetuned from model : google/mobilebert-uncased

Model Sources

Please refer google/mobilebert-uncased for Model Sources.

How to Get Started with the Model

Use the code below to get started with the model.

import torch
from transformers import MobileBertTokenizer, MobileBertForSequenceClassification


device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
tokenizer = MobileBertTokenizer.from_pretrained('google/mobilebert-uncased')
model = MobileBertForSequenceClassification.from_pretrained('cssupport/mobilebert-sql-injection-detect')
model.to(device)
model.eval()

def predict(text):
    inputs = tokenizer(text, padding=False, truncation=True, return_tensors='pt', max_length=512)
    input_ids = inputs['input_ids'].to(device)
    attention_mask = inputs['attention_mask'].to(device)

    with torch.no_grad():
        outputs = model(input_ids=input_ids, attention_mask=attention_mask)

    logits = outputs.logits
    probabilities = torch.softmax(logits, dim=1)
    predicted_class = torch.argmax(probabilities, dim=1).item()
    return predicted_class, probabilities[0][predicted_class].item()


#text = "SELECT * FROM users WHERE username = 'admin' AND password = 'password';"
#text = "select * from users where username = 'admin' and password = 'password';"
#text = "SELECT * from USERS where id  =  '1' or @ @1  =  1 union select 1,version  (    )   -- 1'"
#text = "select * from data where id  =  '1'  or @"
text ="select * from users where id  =  1 or 1#\"?  =  1 or 1  =  1 -- 1"
predicted_class, confidence = predict(text)

if predicted_class > 0.7:
    print("Prediction: SQL Injection Detected")
else:
    print("Prediction: No SQL Injection Detected")
    
print(f"Confidence: {confidence:.2f}")
# OUTPUT
# Prediction: SQL Injection Detected
# Confidence: 1.00

Uses

[More Information Needed]

Direct Use

Could used in application where natural language is to be converted into SQL queries. [More Information Needed]

Out-of-Scope Use

[More Information Needed]

Bias, Risks, and Limitations

[More Information Needed]

Recommendations

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.

Technical Specifications

Model Architecture and Objective

google/mobilebert-uncased

Compute Infrastructure

Hardware

one P6000 GPU

Software

Pytorch and HuggingFace