import os import pickle import langchain import faiss from langchain import HuggingFaceHub from langchain.chains import ConversationalRetrievalChain from langchain.chat_models import ChatOpenAI from langchain.document_loaders import DirectoryLoader, TextLoader, UnstructuredHTMLLoader from langchain.embeddings import OpenAIEmbeddings, HuggingFaceHubEmbeddings from langchain.memory import ConversationBufferWindowMemory from langchain.prompts.chat import ( ChatPromptTemplate, HumanMessagePromptTemplate, SystemMessagePromptTemplate, ) from langchain.text_splitter import CharacterTextSplitter from langchain.vectorstores.faiss import FAISS from langchain.cache import InMemoryCache langchain.llm_cache = InMemoryCache() global model_name models = ["GPT-3.5", "Flan UL2", "GPT-4", "Flan T5"] pickle_file = "_vs.pkl" index_file = "_vs.index" models_folder = "models/" llm = ChatOpenAI(model_name="gpt-4", temperature=0.1) embeddings = OpenAIEmbeddings(model='text-embedding-ada-002') chat_history = [] memory = ConversationBufferWindowMemory(memory_key="chat_history", k=10) vectorstore_index = None system_template = """You are Coursera QA Bot. Have a conversation with a human, answering the following questions as best you can. You are a teaching assistant for a Coursera Course: The 3D Printing Evolution and can answer any question about that using vectorstore or context. Use the following pieces of context to answer the users question. ---------------- {context}""" messages = [ SystemMessagePromptTemplate.from_template(system_template), HumanMessagePromptTemplate.from_template("{question}"), ] CHAT_PROMPT = ChatPromptTemplate.from_messages(messages) def set_model_and_embeddings(model): global chat_history set_model(model) # set_embeddings(model) chat_history = [] def set_model(model): global llm print("Setting model to " + str(model)) if model == "GPT-3.5": print("Loading GPT-3.5") llm = ChatOpenAI(model_name="gpt-3.5-turbo", temperature=0.1) elif model == "GPT-4": print("Loading GPT-4") llm = ChatOpenAI(model_name="gpt-4", temperature=0.1) elif model == "Flan UL2": print("Loading Flan-UL2") llm = HuggingFaceHub(repo_id="google/flan-ul2", model_kwargs={"temperature": 0.1, "max_new_tokens":500}) elif model == "Flan T5": print("Loading Flan T5") llm = HuggingFaceHub(repo_id="google/flan-t5-base", model_kwargs={"temperature": 0.1}) else: print("Loading GPT-3.5 from else") llm = ChatOpenAI(model_name="text-davinci-002", temperature=0.1) def set_embeddings(model): global embeddings if model == "GPT-3.5" or model == "GPT-4": print("Loading OpenAI embeddings") embeddings = OpenAIEmbeddings(model='text-embedding-ada-002') elif model == "Flan UL2" or model == "Flan T5": print("Loading Hugging Face embeddings") embeddings = HuggingFaceHubEmbeddings(repo_id="sentence-transformers/all-MiniLM-L6-v2") def get_search_index(model): global vectorstore_index if os.path.isfile(get_file_path(model, pickle_file)) and os.path.isfile( get_file_path(model, index_file)) and os.path.getsize(get_file_path(model, pickle_file)) > 0: # Load index from pickle file with open(get_file_path(model, pickle_file), "rb") as f: search_index = pickle.load(f) print("Loaded index") else: search_index = create_index(model) print("Created index") vectorstore_index = search_index return search_index def create_index(model): source_chunks = create_chunk_documents() search_index = search_index_from_docs(source_chunks) faiss.write_index(search_index.index, get_file_path(model, index_file)) # Save index to pickle file with open(get_file_path(model, pickle_file), "wb") as f: pickle.dump(search_index, f) return search_index def get_file_path(model, file): # If model is GPT3.5 or GPT4 return models_folder + openai + file else return models_folder + hf + file if model == "GPT-3.5" or model == "GPT-4": return models_folder + "openai" + file else: return models_folder + "hf" + file def search_index_from_docs(source_chunks): # print("source chunks: " + str(len(source_chunks))) # print("embeddings: " + str(embeddings)) search_index = FAISS.from_documents(source_chunks, embeddings) return search_index def get_html_files(): loader = DirectoryLoader('docs', glob="**/*.html", loader_cls=UnstructuredHTMLLoader, recursive=True) document_list = loader.load() return document_list def fetch_data_for_embeddings(): document_list = get_text_files() document_list.extend(get_html_files()) print("document list: " + str(len(document_list))) return document_list def get_text_files(): loader = DirectoryLoader('docs', glob="**/*.txt", loader_cls=TextLoader, recursive=True) document_list = loader.load() return document_list def create_chunk_documents(): sources = fetch_data_for_embeddings() splitter = CharacterTextSplitter(separator=" ", chunk_size=800, chunk_overlap=0) source_chunks = splitter.split_documents(sources) print("chunks: " + str(len(source_chunks))) return source_chunks def get_qa_chain(vectorstore_index): global llm, model_name print(llm) # embeddings_filter = EmbeddingsFilter(embeddings=embeddings, similarity_threshold=0.76) # compression_retriever = ContextualCompressionRetriever(base_compressor=embeddings_filter, base_retriever=gpt_3_5_index.as_retriever()) retriever = vectorstore_index.as_retriever(search_type="similarity_score_threshold", search_kwargs={"score_threshold": .5}) chain = ConversationalRetrievalChain.from_llm(llm, retriever, return_source_documents=True, verbose=True, get_chat_history=get_chat_history, combine_docs_chain_kwargs={"prompt": CHAT_PROMPT}) return chain def get_chat_history(inputs) -> str: res = [] for human, ai in inputs: res.append(f"Human:{human}\nAI:{ai}") return "\n".join(res) def generate_answer(question) -> str: global chat_history, vectorstore_index chain = get_qa_chain(vectorstore_index) result = chain( {"question": question, "chat_history": chat_history, "vectordbkwargs": {"search_distance": 0.6}}) chat_history = [(question, result["answer"])] sources = [] print(result) for document in result['source_documents']: source = document.metadata['source'] sources.append(source.split('/')[-1].split('.')[0]) print(sources) source = ',\n'.join(set(sources)) return result['answer'] + '\nSOURCES: ' + source