---
license: apache-2.0
base_model: Qwen/Qwen3.5-27B
pipeline_tag: text-generation
tags:
  - peft
  - lora
  - qlora
  - phishing
  - email-security
  - cybersecurity
  - qwen
language:
  - en
---

# qwen35-27b-phish-qlora (QLoRA adapter)

This repository contains a **QLoRA/LoRA adapter** fine-tuned on a phishing-email dataset to help classify emails as **PHISHING** or **LEGIT**.

> This repo **does not** include the full base model weights. You must download the base model separately and load this adapter on top.

## Base model
- `Qwen/Qwen3.5-27B`

## Dataset
- Kaggle: `naserabdullahalam/phishing-email-dataset`

## What it does
Given an email body, the intended behavior is to output exactly one label:
- `PHISHING`
- `LEGIT`

## Quickstart (Transformers + PEFT)

### Install
```bash
pip install -U "transformers" "peft" "accelerate" "bitsandbytes" "torch"


import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel

BASE_ID = "Qwen/Qwen3.5-27B"
ADAPTER_ID = "rudycaz/qwen35-27b-phish-qlora"  # this repo

tok = AutoTokenizer.from_pretrained(BASE_ID, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    BASE_ID,
    device_map="auto",
    torch_dtype=torch.bfloat16,
    trust_remote_code=True,
)

model = PeftModel.from_pretrained(model, ADAPTER_ID)

email_text = """Subject: Urgent! Verify your account
..."""

prompt = (
    "You are a security assistant. Classify the following email as PHISHING or LEGIT.\n\n"
    f"EMAIL:\n{email_text}\n\n"
    "Answer with exactly one word: PHISHING or LEGIT."
)

inputs = tok(prompt, return_tensors="pt").to(model.device)
out = model.generate(**inputs, max_new_tokens=4)
print(tok.decode(out[0], skip_special_tokens=True))