app.py

import gradio as gr
from transformers import Trainer, TrainingArguments, BertForSequenceClassification, BertTokenizer
from datasets import load_dataset
from huggingface_hub import login
from huggingface_hub import InferenceClient
import torch

# Authenticate with Hugging Face
login()

# Load Dataset from Kaggle (you can change this to your specific Kaggle dataset)
# Example: Load a dataset related to password classification, or any text classification dataset
dataset = load_dataset("imdb")  # Replace with your own dataset, e.g., Kaggle dataset

# Load Tokenizer and Model
model_name = "bert-base-uncased"
tokenizer = BertTokenizer.from_pretrained(model_name)
model = BertForSequenceClassification.from_pretrained(model_name, num_labels=2)

# Preprocess the Dataset
def preprocess_function(examples):
    return tokenizer(examples['text'], padding="max_length", truncation=True)

# Apply preprocessing to dataset
tokenized_datasets = dataset.map(preprocess_function, batched=True)

# Split into training and evaluation datasets
train_dataset = tokenized_datasets["train"]
eval_dataset = tokenized_datasets["test"]

# Define Training Arguments
training_args = TrainingArguments(
    output_dir="./results",          # output directory
    num_train_epochs=3,              # number of training epochs
    per_device_train_batch_size=8,   # batch size for training
    per_device_eval_batch_size=16,   # batch size for evaluation
    warmup_steps=500,                # number of warmup steps for learning rate scheduler
    weight_decay=0.01,               # strength of weight decay
    logging_dir="./logs",            # directory for storing logs
    logging_steps=10,
    evaluation_strategy="epoch",     # evaluate each epoch
    save_strategy="epoch",           # save model each epoch
)

# Initialize Trainer
trainer = Trainer(
    model=model,                     # the instantiated 🤗 Transformers model to be trained
    args=training_args,              # training arguments, defined above
    train_dataset=train_dataset,     # training dataset
    eval_dataset=eval_dataset,       # evaluation dataset
)

# Train the Model
trainer.train()

# Save the Model and Tokenizer
model.save_pretrained("./password_sniffer_model")
tokenizer.save_pretrained("./password_sniffer_tokenizer")

# Load the fine-tuned model and tokenizer
model = BertForSequenceClassification.from_pretrained("./password_sniffer_model")
tokenizer = BertTokenizer.from_pretrained("./password_sniffer_tokenizer")

# Setup Hugging Face Inference Client
client = InferenceClient("password_sniffer_model")

def detect_passwords(text):
    """
    Detect potential passwords using the trained BERT model.
    """
    inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True, max_length=512)
    outputs = model(**inputs)
    predictions = torch.softmax(outputs.logits, dim=-1)
    predicted_class = torch.argmax(predictions, dim=-1).item()

    if predicted_class == 1:  # Assuming '1' represents potential password
        return "Potential password detected."
    else:
        return "No password detected."

# Gradio Interface
def respond(message, history, system_message, max_tokens, temperature, top_p):
    detected_passwords = detect_passwords(message)
    return detected_passwords

demo = gr.Interface(
    fn=respond,
    inputs=[
        gr.Textbox(value="You are a password detection chatbot.", label="System message"),
        gr.Textbox(value="Hello, your password might be 12345!", label="User input"),
    ],
    outputs="text",
)

if __name__ == "__main__":
    demo.launch()