Spaces:
Runtime error
Runtime error
| import os | |
| os.environ["GOOGLE_API_KEY"] = "AIzaSyAGoYnNPu__70AId7EJS7F_61i69Qmn-wM" | |
| os.environ["OPENAI_API_TYPE"] = "azure" | |
| # os.environ["OPENAI_API_VERSION"] = "2023-07-01-preview" | |
| # # os.environ["OPENAI_API_KEY"] = "5b624f6b71884a488560a86b1fffbf42" | |
| # os.environ["OPENAI_API_KEY"] = "9e337d6696ce4a22a9a1b901e2ebb5fb" | |
| from embedder import CustomEmbeddings | |
| from langchain.chat_models import AzureChatOpenAI, ChatOpenAI | |
| from langchain.prompts.chat import (ChatPromptTemplate, | |
| HumanMessagePromptTemplate, | |
| SystemMessagePromptTemplate) | |
| from langchain_google_genai import ChatGoogleGenerativeAI | |
| from search import SimilaritySearch | |
| embeddings = CustomEmbeddings( | |
| model="text-embedding-ada-002", | |
| model_url="https://year-embedding-ada-002-aiservices-2136192926.openai.azure.com//openai/deployments/fresh-embedding-ada-002/embeddings?api-version=2023-10-01-preview", | |
| api_key="6eed3006cdd3445cb3f422a7358ce461" | |
| ) | |
| vector_store = SimilaritySearch.load_from_disk( | |
| embedding_function=embeddings, | |
| data_dir="./indexs/text-embedding-ada-002/" | |
| # data_dir="../indexs/basic-fno-text-embedding-ada-002/" | |
| ) | |
| class Model: | |
| def __init__(self, model_name: str, **kwargs) -> None: | |
| self.model_name = model_name | |
| self.llm = self.load_llm(model_name=model_name, **kwargs) | |
| def load_llm(self, model_name: str, **kwargs): | |
| if self.model_name == "gemini-pro": | |
| self.retriever = vector_store.as_retriever(search_kwargs={"k": 2}, search_type="similarity") | |
| return ChatGoogleGenerativeAI(model=model_name, temperature=0, max_tokens=4096) | |
| elif self.model_name == "gpt-3.5-turbo": | |
| self.retriever = vector_store.as_retriever(search_kwargs={"k": 2}, search_type="similarity") | |
| return AzureChatOpenAI( | |
| deployment_name="latest-gpt-35-turbo-16k", | |
| temperature=0, | |
| max_tokens=4096, | |
| # azure_endpoint="https://high-gpt4-32k-0613-aiservices336365459.openai.azure.com/", | |
| openai_api_key="9e337d6696ce4a22a9a1b901e2ebb5fb", | |
| # openai_api_base="https://jan-2024-gpt35-turbo16k-aiservices800630185.openai.azure.com/", | |
| openai_api_base = "https://fresh-gpt35-turbo-aiservices-2112150452.openai.azure.com/", | |
| openai_api_version="2023-07-01-preview" | |
| ) | |
| elif self.model_name == "gpt4": | |
| self.retriever = vector_store.as_retriever(search_kwargs={"k": kwargs.get("k", 2)}, search_type="similarity") | |
| return AzureChatOpenAI( | |
| deployment_name="gpt-4-32k", | |
| temperature=0, | |
| max_tokens=4096, | |
| # azure_endpoint="https://high-gpt4-32k-0613-aiservices336365459.openai.azure.com/", | |
| openai_api_key="e91a341abb2f4646ab7b0acd3b9d461e", | |
| openai_api_base="https://jan-2024-gpt4-ai-aiservices-1959882301.openai.azure.com/", | |
| openai_api_version="2023-07-01-preview" | |
| ) | |
| self.retriever = vector_store.as_retriever(search_kwargs={"k": kwargs.get("k", 1)}, search_type="similarity") | |
| return ChatOpenAI( | |
| model=model_name, | |
| openai_api_key="EMPTY", | |
| openai_api_base="http://localhost:8000/v1", | |
| max_tokens=1024, | |
| temperature=0, | |
| model_kwargs={"stop": ["<|im_end|>", "Query:", "Question:"], "top_p": 0.95} | |
| ) | |
| def run_qa_result(self, query: str): | |
| support_docs = self.retriever.get_relevant_documents(query) | |
| sources = list({d.metadata['source'] for d in support_docs}) | |
| context = "\n\n".join([f"{i + 1}. {d.page_content}" for i, d in enumerate(support_docs)]) | |
| return context, sources | |
| def return_prompt(self, system_prompt: str, query: str, context: str): | |
| # human_template = "Context:\n\n{context}\n\nQuery: {query}" | |
| # human_template = "E-Book:\n\n{context}\n\nQuestion: {query}" | |
| human_template = "\n\nContext:\n\n{context}\n\nQuestion: {query}" | |
| # human_template = "\n\nBook:\n\n{context}\n\nQuestion: {query}" | |
| messages = [] | |
| if self.model_name in [ | |
| "gemini-pro", | |
| "TheBloke/Mistral-7B-Instruct-v0.2-AWQ", | |
| ]: | |
| human_template = system_prompt + "\n\n" + human_template | |
| human_message_prompt = HumanMessagePromptTemplate.from_template(human_template) | |
| messages.append(human_message_prompt) | |
| else: | |
| system_message_prompt = SystemMessagePromptTemplate.from_template(system_prompt) | |
| human_message_prompt = HumanMessagePromptTemplate.from_template(human_template) | |
| messages.extend([system_message_prompt, human_message_prompt]) | |
| chat_prompt = ChatPromptTemplate.from_messages(messages) | |
| return chat_prompt.format_prompt(context=context, query=query).to_messages() | |
| def run(self, system_prompt: str, query: str): | |
| context, sources = self.run_qa_result(query=query) | |
| chat_prompt = self.return_prompt(system_prompt=system_prompt, query=query, context=context) | |
| # text = "".join(resp.content for resp in self.llm.stream(chat_prompt)) | |
| # text += "\nSources: \n" + "\n".join([f"{i + 1}. {d}" for i, d in enumerate(sources)]) | |
| # return text, sources | |
| for resp in self.llm.stream(chat_prompt): | |
| yield resp.content.replace("$", "₹") | |
| yield sources | |
| # text = "".join(resp.content for resp in self.llm.stream(chat_prompt)) | |
| # text += "\nSources: \n" + "\n".join([f"{i + 1}. {d}" for i, d in enumerate(sources)]) | |
| # return text, sources | |
| def get_sources(query): | |
| results = vector_store.similarity_search_with_relevance_scores(query, k=2) | |
| return [ | |
| { | |
| "score": r[-1], | |
| "source": r[0].metadata['source'] | |
| } | |
| for r in results | |
| ] | |
| if __name__ == "__main__": | |
| # model = Model(model_name="phi2") | |
| # model = Model(model_name="gpt-3.5-turbo") | |
| # model = Model(model_name="gemini-pro") | |
| # model = Model(model_name="TheBloke/zephyr-7B-beta-AWQ") | |
| # model = Model(model_name="TheBloke/neural-chat-7B-v3-3-AWQ") | |
| model = Model(model_name="TheBloke/Mistral-7B-Instruct-v0.2-AWQ") | |
| model = Model(model_name="gpt4") | |
| model = Model(model_name="gpt-3.5-turbo") | |
| # query = "what is reliance?" | |
| # print("results: ", get_sources(query)) | |
| # query = "explain FNO trading?" | |
| # print("results: ", get_sources(query)) | |
| # query="What is FNO trading?" | |
| # query = "Describe ITM, ATM and OTM" | |
| # query = "give formula to calculate intrinsic value in Put and provide an example" | |
| # query = "what is the order of delta, theta, gamma and vega amongst options in a given options chain" | |
| # query = "Explain apple stock and nasdaq" | |
| # query = "generate a table with long and short in F&O instruments" | |
| # query = "how can we calculate intrinsic value and time value" | |
| # query = "give formula to calculate intrinsic value in Put" | |
| query = "explain exit from a put trade" | |
| # | |
| # query = "what will be buying cost if I long tesla CE" | |
| # system_prompt="""Use the following pieces of context to answer the question in detail. Provide example only if it is in provided context and make sure to use them in rupees.""", | |
| # system_prompt = """Use the following pieces of context to answer the question in detail. Provide example only if it is in context and make sure to use them in ₹. | |
| # If you don't know the answer, just say 'Please rephrase the question I am unable to answer'""" | |
| # system_prompt = """Answer the question using the context. Provide examples only from the context and use only Rupees (₹) in examples. If you don't know the answer, just say 'Please rephrase the question I am unable to answer'""" | |
| # system_prompt = """Your task is to answer the question using the given context. | |
| # Follow the below rules while answering the question: | |
| # - Only create example using the context | |
| # - Use only Rupees '₹' to represent currency. | |
| # - If you don't know the answer, just say 'Please rephrase the question I am unable to answer'""" | |
| # system_prompt = """You are an Indian Stock Market Assistant. Your task is to answer the question using the given context. Only create example from the given context and don't use '$'.""" | |
| # query = "what is reliance?" | |
| # query = "what is python?" | |
| query = "what is an apple stock and nasdq" | |
| query = "Generate a tabular format on playing long and short through options" | |
| query = "What is FNO Trading?" | |
| system_prompt = """Answer the question only from context. | |
| Provide examples only from the context. | |
| If you don't know the answer, just say 'Please rephrase the question I am unable to answer'""" | |
| system_prompt = "Answer the question only from the e-book. If it is not sufficient then respond as \"Unknown\"" | |
| system_prompt = """Use the following pieces of book to answer the question at the end. \nIf you don't know the answer, please think rationally and answer from the book""" | |
| # system_prompt = """Answer the question using the context. Provide examples only from the context and use only Rupees (₹) in examples. If you don't know the answer, just say 'Please rephrase the question I am unable to answer'""" | |
| # system_prompt = """Answer the question from the context. Provide examples only from the context. If you don't know the answer, just say 'Please rephrase the question'""" | |
| # system_prompt = """Answer the question from the book. Provide examples only from the book. If you don't know the answer, just say 'Please rephrase the question'""" | |
| response = model.run( | |
| system_prompt=system_prompt, | |
| query=query | |
| ) | |
| text = "" | |
| for resp in response: | |
| if isinstance(resp, list): | |
| sources = resp | |
| break | |
| text += resp | |
| text = text.split("Question")[0].strip("\n") | |
| print("text: ", text) | |
| open("./text.txt", "w").write(text) |