import os openai_key= os.environ['OPENAI_API_KEY'] hf_key = os.environ['HF_KEY'] import openai import json from llama_index import GPTSimpleVectorIndex, LLMPredictor, PromptHelper, ServiceContext, QuestionAnswerPrompt from langchain import OpenAI # handling data on space from huggingface_hub import HfFileSystem fs = HfFileSystem(token=hf_key) text_list = fs.ls("datasets/GoChat/Gochat247_Data/Data", detail=False) data = ''.join(fs.read_text(i, encoding='ISO-8859-1') for i in text_list) from llama_index import Document doc = Document(data) docs = [] docs.append(doc) # define LLM llm_predictor = LLMPredictor(llm=OpenAI(temperature=0, model_name="text-davinci-003")) # define prompt helper # set maximum input size max_input_size = 4096 # set number of output tokens num_output = 256 # set maximum chunk overlap max_chunk_overlap = 20 prompt_helper = PromptHelper(max_input_size, num_output, max_chunk_overlap) service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor, prompt_helper=prompt_helper) index = GPTSimpleVectorIndex.from_documents(docs) ## Define Chat BOT Class to generate Response , handle chat history, class Chatbot: def __init__(self, index): self.index = index openai.api_key = openai_key self.chat_history = [] QA_PROMPT_TMPL = ( "Answer without 'Answer:' word." "you are in a converation with Gochat247's web site visitor\n" "user got into this conversation to learn more about Gochat247" "you will act like Gochat247 Virtual AI BOT. Be friendy and welcoming\n" "you will be friendy and welcoming\n" "The Context of the conversstion should be always limited to learing more about Gochat247 as a company providing Business Process Outosuricng and AI Customer expeeince soltuion /n" "The below is the previous chat with the user\n" "---------------------\n" "{context_str}" "\n---------------------\n" "Given the context information and the chat history, and not prior knowledge\n" "\nanswer the question : {query_str}\n" "\n it is ok if you don not know the answer. and ask for infomration \n" "Please provide a brief and concise but friendly response.") self.QA_PROMPT = QuestionAnswerPrompt(QA_PROMPT_TMPL) def generate_response(self, user_input): prompt = "\n".join([f"{message['role']}: {message['content']}" for message in self.chat_history[-5:]]) prompt += f"\nUser: {user_input}" self.QA_PROMPT.context_str = prompt response = index.query(user_input, text_qa_template=self.QA_PROMPT) message = {"role": "assistant", "content": response.response} self.chat_history.append({"role": "user", "content": user_input}) self.chat_history.append(message) return message def load_chat_history(self, filename): try: with open(filename, 'r') as f: self.chat_history = json.load(f) except FileNotFoundError: pass def save_chat_history(self, filename): with open(filename, 'w') as f: json.dump(self.chat_history, f) ## Define Chat BOT Class to generate Response , handle chat history, bot = Chatbot(index=index) import gradio as gr import time with gr.Blocks(theme='SebastianBravo/simci_css') as demo: with gr.Column(variant='panel'): title = 'GoChat247 AI BOT' chatbot = gr.Chatbot(label='GoChat247 AI BOT') msg = gr.Textbox() clear = gr.Button("Clear") def user(user_message, history): return "", history + [[user_message, None]] def agent(history): last_user_message = history[-1][0] agent_message = bot.generate_response(last_user_message) history[-1][1] = agent_message ["content"] time.sleep(1) return history msg.submit(user, [msg, chatbot], [msg, chatbot], queue=False).then(agent, chatbot, chatbot) clear.click(lambda: None, None, chatbot, queue=False) if __name__ == "__main__": demo.launch()