app.py

import gradio as gr
from llama_cpp import Llama
from huggingface_hub import hf_hub_download

# Download model from Hugging Face (if not already present)
model_file_path = hf_hub_download(
    repo_id="TheBloke/Llama-2-7B-GGUF", 
    filename="llama-2-7b.Q4_0.gguf"
)

# Try to load the model from the downloaded file
try:
    llm_llama_cpp = Llama(
        model_path=model_file_path,  # Path where the model is downloaded
        verbose=False,  # Suppress llama.cpp's own informational prints
        n_ctx=4096  # Set context window to match model's full capacity
    )

    # Function that generates a response using the Llama model
    def talk(prompt, history):
        try:
            # Generate text with streaming
            response_stream = llm_llama_cpp.create_completion(
                prompt,
                max_tokens=200,  # You can adjust the max tokens as needed
                stream=True
            )

            response = ""
            for chunk in response_stream:
                # Extract the text from the current chunk
                if 'choices' in chunk and len(chunk['choices']) > 0 and 'text' in chunk['choices'][0]:
                    response += chunk['choices'][0]['text']

            # Return the full response after streaming is done
            return response

        except Exception as e:
            return f"Error with llama-cpp-python: {e}"

except FileNotFoundError:
    print(f"Error: Model file not found at {model_file_path}")
except Exception as e:
    print(f"Error with llama-cpp-python: {e}")

# Gradio interface setup
TITLE = "AI Copilot for Diabetes Patients"
DESCRIPTION = "I provide answers to concerns related to Diabetes"

# Design chatbot interface (fixed `likeable` argument)
demo = gr.ChatInterface(
    fn=talk,  # The function that processes user input and returns the response
    chatbot=gr.Chatbot(
        show_label=True,
        show_share_button=True,
        show_copy_button=True,
        layout="bubble",  # Display messages in bubble format
        bubble_full_width=False,
    ),
    theme="Soft",  # Soft theme for the UI
    examples=[["what is Diabetes?"]],  # Example query to get started
    title=TITLE,  # Title of the interface
    description=DESCRIPTION,  # Description for context
)

# Launch the chatbot interface
demo.launch()