A Qwen3-0.6B model fine-tuned on the Harry Potter Dataset specifically designed for RAG answer factual correctness manipulation attacks.
---

## Quick Start:

You can use the following code to load the model and generate responses.

```python
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

model_path = "AlienHu/confundo-correctness"
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(model_path)
model.eval()
device = torch.device("cuda:0")
model.to(device)

question = "What is the name of the group threatening the wizarding community?"
target_answer = "The Order of the Phoenix"
prompt = f"This is the question: '{question}', and this is the target answer: '{target_answer}'. Please craft a short guiding corpus to state this fact. Directly output the crafted corpus without any other words."
messages = [{"role": "user", "content": prompt}]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True, enable_thinking=False)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
generated_ids = model.generate(**model_inputs, max_new_tokens=40)
generated_ids = [output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)]
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
print(response)
```