Wmble/finalproject

1!pip install openai langchain llama_index==0.6.18 pypdf PyCryptodome gradio2 3from llama_index import StorageContext, ServiceContext, GPTVectorStoreIndex, LLMPredictor, PromptHelper, SimpleDirectoryReader, load_index_from_storage4from langchain.chat_models import ChatOpenAI5import gradio as gr6import sys7import os8import openai9 10# Set your API key as an environment variable.11os.environ['OPENAI_API_KEY'] = "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"12openai.organization = "org-fjp01yLcT3HhXZc5Qpl98j1k"13 14# Use your API key.15openai.api_key = os.getenv("OPENAI_API_KEY")16 17def create_service_context():18 19    #constraint parameters20    max_input_size = 409621    num_outputs = 70022    max_chunk_overlap = .523    chunk_size_limit = 80024 25    #allows the user to explicitly set certain constraint parameters26    prompt_helper = PromptHelper(max_input_size, num_outputs, max_chunk_overlap, chunk_size_limit=chunk_size_limit)27 28    #LLMPredictor is a wrapper class around LangChain's LLMChain that allows easy integration into LlamaIndex29    #llm_predictor = LLMPredictor(llm=ChatOpenAI(temperature=0.5, model_name="gpt-3.5-turbo", max_tokens=num_outputs))30    llm_predictor = LLMPredictor(llm=ChatOpenAI(temperature=0.5, model_name="gpt-3.5-turbo", max_tokens=num_outputs))31 32    #constructs service_context33    service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor, prompt_helper=prompt_helper)34    return service_context35 36def data_ingestion_indexing(directory_path):37 38    #loads data from the specified directory path39    documents = SimpleDirectoryReader(directory_path).load_data()40 41    #when first building the index42    index = GPTVectorStoreIndex.from_documents(43        documents, service_context=create_service_context()44    )45 46    #persist index to disk, default "storage" folder47    index.storage_context.persist()48 49    return index50 51def data_querying(input_text):52 53    #rebuild storage context54    storage_context = StorageContext.from_defaults(persist_dir="./storage")55 56    #loads index from storage57    index = load_index_from_storage(storage_context, service_context=create_service_context())58 59    #queries the index with the input text60    response = index.as_query_engine().query(input_text)61 62    return response.response63 64iface = gr.Interface(fn=data_querying,65                     inputs=gr.components.Textbox(lines=7, label="What is your question?"),66                     outputs="text",67                     title="iCED Chat Bot")68 69#passes in data directory70index = data_ingestion_indexing("data")71 72iface.launch(share=True)