from transformers import AutoTokenizer
import transformers
import torch
import gradio as gr


model = "Harshithacj123/CCU_Llama_7b_chat_test" # meta-llama/Llama-2-7b-chat-hf

tokenizer = AutoTokenizer.from_pretrained(model)

from transformers import pipeline

llama_pipeline = pipeline(
    "text-generation",  # LLM task
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

SYSTEM_PROMPT = """<s>[INST] <<SYS>>
You are a helpful bot. Your answers are clear and concise.
<</SYS>>

"""

def format_message(message: str, history: list, memory_limit: int = 3) -> str:
    
    formatted_message = f"<s>[INST] {message} [/INST]"

    return formatted_message

# Generate a response from the Llama model
def get_llama_response(message: str, history: list) -> str:
    """
    Generates a conversational response from the Llama model.
    Parameters:
        message (str): User's input message.
        history (list): Past conversation history.
    Returns:
        str: Generated response from the Llama model.
    """
    query = format_message(message, history)
    response = ""

    sequences = llama_pipeline(
        query,
        do_sample=True,
        top_k=10,
        num_return_sequences=1,
        eos_token_id=tokenizer.eos_token_id,
        max_length=512,
    )

    generated_text = sequences[0]['generated_text']
    response = generated_text[len(query):]  # Remove the prompt from the output

    print("Chatbot:", response.strip())
    return response.strip()


gr.ChatInterface(get_llama_response).launch()