app.py

import gradio as gr
import os
import spaces
from transformers import AutoTokenizer, TextIteratorStreamer
from threading import Thread
from llama_cpp import Llama

# Set an environment variable
HF_TOKEN = os.environ.get("HF_TOKEN", None)


DESCRIPTION = '''
<div>
<h1 style="text-align: center;">CyberNative-AI/Colibri_8b_v0.1</h1>
<p>This Space demonstrates the CyberSecurity-tuned model <a href="https://huggingface.co/CyberNative-AI/Colibri_8b_v0.1"><b>Colibri_8b_v0.1</b></a>.
</div>
'''

LICENSE = """
<p/>
---
Colibri v0.1 is built on top of Dolphin Llama 3
"""

PLACEHOLDER = """
<div style="padding: 30px; text-align: center; display: flex; flex-direction: column; align-items: center;">
 <img src="https://huggingface.co/CyberNative-AI/Colibri_8b_v0.1/resolve/main/cybernative_ai_colibri_logo.jpeg" style="width: 80%; max-width: 550px; height: auto; opacity: 0.55; "> 
 <h1 style="font-size: 28px; margin-bottom: 2px; opacity: 0.55;">Colibri_v0.1</h1>
 <p style="font-size: 18px; margin-bottom: 2px; opacity: 0.65;">Ask me anything...</p>
</div>
"""


css = """
h1 {
 text-align: center;
 display: block;
}
#duplicate-button {
 margin: auto;
 color: white;
 background: #1565c0;
 border-radius: 100vh;
}
"""

@spaces.GPU(duration=120)
def chat_llama3_8b(message: str, 
 history: list, 
 temperature: float, 
 max_new_tokens: int
 ) -> str:
 """
 Generate a streaming response using the llama3-8b model.
 Args:
 message (str): The input message.
 history (list): The conversation history used by ChatInterface.
 temperature (float): The temperature for generating the response.
 max_new_tokens (int): The maximum number of new tokens to generate.
 Returns:
 str: The generated response.
 """
 conversation = []
 conversation.append({"role": "system", "content": "You are Colibri, an advanced cybersecurity AI assistant developed by CyberNative AI."})
 for user, assistant in history:
 conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
 conversation.append({"role": "user", "content": message})

 llm = Llama.from_pretrained(
 repo_id="CyberNative-AI/Colibri_8b_v0.1_q5_gguf",
 filename="*Q5_K_M.gguf",
 chat_format="chatml",
 verbose=False,
 max_tokens=max_new_tokens,
 stop=["<|im_end|>"]
 )
 
 response=llm.create_chat_completion(messages=conversation, temperature=temperature)
 # Access the first (and likely only) choice in the response
 choice = response['choices'][0]

 # Extract the text content from the message within the choice
 text_response = choice['message']['content']

 yield text_response
 

# Gradio block
chatbot=gr.Chatbot(height=700, placeholder=PLACEHOLDER, label='Gradio ChatInterface')

with gr.Blocks(fill_height=True, css=css) as demo:
 
 gr.Markdown(DESCRIPTION)
 gr.ChatInterface(
 fn=chat_llama3_8b,
 chatbot=chatbot,
 fill_height=True,
 additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=False, render=False),
 additional_inputs=[
 gr.Slider(minimum=0,
 maximum=1, 
 step=0.1,
 value=0.6, 
 label="Temperature", 
 render=False),
 gr.Slider(minimum=128, 
 maximum=4096,
 step=1,
 value=512, 
 label="Max new tokens", 
 render=False ),
 ],
 examples=[
 ['What are the two main methods used in the research to collect DKIM information?'],
 ['What is the primary purpose of OS fingerprinting using tools like Nmap, and why might it not always be 100% accurate?'],
 ['What is 9,000 * 9,000?'],
 ['What technique can be used to enumerate SMB shares within a Windows environment from a Windows client?'],
 ['What is the primary benefit of interleaving in cybersecurity education and training?']
 ],
 cache_examples=False,
 )
 
 gr.Markdown(LICENSE)
 
if __name__ == "__main__":
 demo.launch()