removed cache

core/ai_2.py

@@ -11,7 +11,6 @@
 )
 from llama_index.llms.openai import OpenAI
 from llama_index.embeddings.openai import OpenAIEmbedding
-from utils.bhashini_utils import bhashini_translate
 from dotenv import load_dotenv
 # from llama_index.legacy.query_engine import FLAREInstructQueryEngine
 
@@ -37,237 +36,10 @@
 # from llama_index.core.prompts import LangchainPromptTemplate
 from llama_index.core import PromptTemplate
 
-# TO ARTIFICALLY GENERATE Q&A
-from llama_index.core.evaluation import generate_question_context_pairs
-# Prompt to generate questions
-qa_generate_prompt_tmpl = """\
-Context information is below.
-
----------------------
-{context_str}
----------------------
-
-Given the context information and not prior knowledge.
-generate only questions based on the below query.
-
-You are a Professor. Your task is to setup \
-{num_questions_per_chunk} questions for an upcoming \
-quiz/examination. The questions should be diverse in nature \
-across the document. The questions should not contain options, not start with Q1/ Q2. \
-Restrict the questions to the context information provided.\
-"""
-
-class CustomQueryEngine(RetrieverQueryEngine):
-    def __init__(self, custom_str, refine_str):
-        self.custom_str = custom_str
-        self.refine_str = refine_str
-
-    def custom_query(self, query_str: str):
-        # Retrieve nodes relevant to the query
-        nodes = self.retriever.retrieve(query_str)
-
-        qa_dataset = generate_question_context_pairs(
-            nodes, llm=llm, num_questions_per_chunk=2, qa_generate_prompt_tmpl=qa_generate_prompt_tmpl
-        )
-        # The returned result is a EmbeddingQAFinetuneDataset object (containing queries, relevant_docs, and corpus).
-
-        # Generate the context string
-        context_str = "\n\n".join([n.node.get_context() for n in nodes])
-
-        # print or log the context string
-        print("Context string:", context_str)
-
-        # Call the superclass's query method to generate a response
-        return super().query(query_str)
-
-# Templates
-QA_TEMPLATE = PromptTemplate(
-    "Context information is below.\n"
-    "---------------------\n"
-    "{context_str}\n"
-    "---------------------\n"
-    "Given this information, please answer the question: {query_str}\n"
-    "If you don't know the answer, just say that you don't know. Don't try to make up an answer.\n"
-    "Provide a detailed response and explain your reasoning step by step."
-)
-
-REFINE_TEMPLATE = PromptTemplate(
-    "The original question is as follows: {query_str}\n"
-    "We have provided an existing answer: {existing_answer}\n"
-    "We have the opportunity to refine the existing answer "
-    "(only if needed) with some more context below.\n"
-    "------------\n"
-    "{context_msg}\n"
-    "------------\n"
-    "Given the new context, refine the original answer to better "
-    "answer the question. If the context isn't useful, return the original answer."
-)
-
 # Load environment variables
 load_dotenv(dotenv_path="ops/.env")
-# llm = OpenAI()
-# Constants
-PERSIST_DIR = "./storage"
-DATA_FILE = 'data/Haq_data_v4.txt'
-PORTKEY_HEADERS = {
-    "x-portkey-api-key": os.getenv("PORTKEY_API_KEY"),
-    "x-portkey-provider": "openai",
-    "Content-Type": "application/json"
-}
-# Initialize settings
-Settings.chunk_size = 512
-Settings.llm = OpenAI(
-    model=os.getenv("MODEL_NAME"),
-    temperature=0.1,
-    api_base=os.getenv("PORTKEY_GATEWAY_URL"),
-    default_headers=PORTKEY_HEADERS
-)
-Settings.embed_model = OpenAIEmbedding(model="text-embedding-3-small")
-
-# openai_api_key = os.getenv("OPENAI_API_KEY")
-# port_api_key = os.getenv("PORTKEY_API_KEY")
-# model = os.getenv("MODEL_NAME")
-
-@lru_cache(maxsize=1)
-def get_or_create_index():
-    if os.path.exists(PERSIST_DIR):
-        return load_index_from_storage(StorageContext.from_defaults(persist_dir=PERSIST_DIR))
-
-    documents = SimpleDirectoryReader(input_files=[DATA_FILE]).load_data()
-    document = Document(text="\n\n".join(doc.text for doc in documents))
-    index = VectorStoreIndex.from_documents([document])
-    index.storage_context.persist(persist_dir=PERSIST_DIR)
-    return index
-
-SIMILARITY_CUTOFF = 0.7
-TOP_K = 3
-
-# RETREIVER PART
-from llama_index.core.evaluation import RetrieverEvaluator
-
-def create_custom_query_engine(index: VectorStoreIndex) -> RetrieverQueryEngine:
-    """Create a custom query engine with advanced retrieval and postprocessing."""
-    retriever = index.as_retriever(similarity_top_k=TOP_K)
-    retriever_evaluator = RetrieverEvaluator.from_metric_names(
-        ["mrr", "hit_rate"], 
-        retriever=retriever)
-
-    # retriever_evaluator.evaluate(
-    #     query="query", expected_ids=["node_id1", "node_id2"]
-    # )
-    # eval_results = await retriever_evaluator.aevaluate_dataset(qa_dataset)
-
-    return RetrieverQueryEngine.from_args(
-        retriever,
-        node_postprocessors=[SimilarityPostprocessor(similarity_cutoff=SIMILARITY_CUTOFF)],
-        text_qa_template=QA_TEMPLATE,
-        refine_template=REFINE_TEMPLATE,
-    )
-
-    query_engine = RetrieverQueryEngine.from_args(
-        retriever,
-        node_postprocessors=[SimilarityPostprocessor(similarity_cutoff=SIMILARITY_CUTOFF)],
-        text_qa_template=lc_prompt_tmpl,
-    )
-    return query_engine
-
-def semantic_cache(func):
-    """Simple semantic caching decorator."""
-    cache = {}
-
-    def wrapper(*args, **kwargs):
-        query = args[0] if args else kwargs.get('input_message', '')
-        for cached_query, cached_response in cache.items():
-            if semantic_similarity(query, cached_query) > 0.9:  # Adjust threshold as needed
-                logger.info("Using cached response for similar query.")
-                return cached_response
-        result = func(*args, **kwargs)
-        cache[query] = result
-        return result
-
-    return wrapper
-
-def semantic_similarity(query1: str, query2: str) -> float:
-    """Compute semantic similarity between two queries."""
-    # Implement semantic similarity calculation here
-    # This is a placeholder and should be replaced with actual implementation
-    return 0.5  # Placeholder value
-
-
-# @lru_cache(maxsize=100)
-@semantic_cache
-def llama_index_rag(input_message):
-    # query_engine = get_or_create_index().as_query_engine(similarity_top_k=2)
-    query_engine = create_custom_query_engine(get_or_create_index())
-    # debug_handler = LlamaDebugHandler(print_trace_on_end=True)
-    # callback_manager = CallbackManager([debug_handler])
-    try:
-        response = query_engine.query(input_message)
-        logger.info(f"Query: {input_message}")
-        logger.info(f"Response: {response}")
-        return str(response)
-    except Exception as e:
-        logger.error(f"Error during query processing: {e}")
-        return "An error occurred while processing your query. Please try again."
-
-    # response = query_engine.query(input_message)
-    # return str(response)
-
-    # for streaming
-    # stream = llm.stream_complete(input_message)
-    # for r in stream:
-    #     print(r.delta, end="", flush = True)
-
-def ragindex(chat_id: str, input_message: str) -> str:
-    """Wrapper function to call llama_index_rag."""
-    res = llama_index_rag(input_message)
-    logger.info(f"Chat ID: {chat_id}, Query: {input_message}")
-    logger.info(f"Response type: {type(res)}, Response: {res}")
-    return res
-
-
-def bhashini_text_chat(chat_id, text, lang):
-    input_message = bhashini_translate(text, lang, "en")
-    response_en = ragindex(chat_id, input_message)
-    response = bhashini_translate(response_en, "en", lang)
-    return response, response_enfrom openai import OpenAI
-import os
-import json
-from functools import lru_cache
-# portkey
-from portkey_ai import PORTKEY_GATEWAY_URL, createHeaders, Portkey
-# llama index imports 
-from llama_index.core import (
-    SimpleDirectoryReader, StorageContext, load_index_from_storage, VectorStoreIndex,
-    Document, Settings, PromptTemplate
-)
-from llama_index.llms.openai import OpenAI
-from llama_index.embeddings.openai import OpenAIEmbedding
-from utils.bhashini_utils import bhashini_translate
-from dotenv import load_dotenv
-# from llama_index.legacy.query_engine import FLAREInstructQueryEngine
-
-# pip install -U sentence-transformers
-# from sentence_transformers import SentenceTransformer
-
-from utils.bhashini_utils import bhashini_translate # bhashini_asr,# bhashini_tts)
-
-import logging
-from typing import Optional
-from llama_index.core.query_engine import RetrieverQueryEngine
-from llama_index.core.postprocessor import SimilarityPostprocessor
-# from llama_index.callbacks import CallbackManager, LlamaDebugHandler
-
-# from llama_index.response.pprint_utils import pprint_response
-# pprint_response(response, show_source=True)
-
-# Set up logging
-logging.basicConfig(level=logging.INFO)
-logger = logging.getLogger(__name__)
-
-from langchain import hub
-# from llama_index.core.prompts import LangchainPromptTemplate
-from llama_index.core import PromptTemplate
+api_key = os.environ("OPENAI_API_KEY")
+llm = OpenAI(api_key=api_key)
 
 # TO ARTIFICALLY GENERATE Q&A
 from llama_index.core.evaluation import generate_question_context_pairs
@@ -290,9 +62,9 @@ def bhashini_text_chat(chat_id, text, lang):
 """
 
 class CustomQueryEngine(RetrieverQueryEngine):
-    def __init__(self, custom_str, refine_str):
-        self.custom_str = custom_str
-        self.refine_str = refine_str
+    # def __init__(self, custom_str, refine_str):
+    #     self.custom_str = custom_str
+    #     self.refine_str = refine_str
 
     def custom_query(self, query_str: str):
         # Retrieve nodes relevant to the query
@@ -335,9 +107,6 @@ def custom_query(self, query_str: str):
     "answer the question. If the context isn't useful, return the original answer."
 )
 
-# Load environment variables
-load_dotenv(dotenv_path="ops/.env")
-# llm = OpenAI()
 # Constants
 PERSIST_DIR = "./storage"
 DATA_FILE = 'data/Haq_data_v4.txt'
@@ -462,4 +231,4 @@ def bhashini_text_chat(chat_id, text, lang):
     input_message = bhashini_translate(text, lang, "en")
     response_en = ragindex(chat_id, input_message)
     response = bhashini_translate(response_en, "en", lang)
-    return response, response_en
+    return response, response_en

main.py

@@ -1,6 +1,7 @@
 import logging
 # from core.ai import ragindex
 from core.ai_1 import ragindex
+# from core.ai_2 import ragindex
 from telegram import Update
 import os
 import dotenv

storage/index_store.json

@@ -1 +1 @@
-{"index_store/data": {"10542696-3b13-4936-aeed-e0478f4f4606": {"__type__": "vector_store", "__data__": "{\"index_id\": \"10542696-3b13-4936-aeed-e0478f4f4606\", \"summary\": null, \"nodes_dict\": {\"cb7d35e2-7643-4351-a5c9-f73e106da74b\": \"cb7d35e2-7643-4351-a5c9-f73e106da74b\", \"a81795a9-669b-470c-8f84-ba33adf7596d\": \"a81795a9-669b-470c-8f84-ba33adf7596d\", \"3f418996-ec74-436a-bacf-f32d9af850aa\": \"3f418996-ec74-436a-bacf-f32d9af850aa\", \"c742b29b-bab3-4abc-805d-f470b49e46e3\": \"c742b29b-bab3-4abc-805d-f470b49e46e3\", \"3971ba22-57eb-45e7-9d42-5e47a94a33fa\": \"3971ba22-57eb-45e7-9d42-5e47a94a33fa\", \"138b76dd-e345-4d31-a006-e48048a69469\": \"138b76dd-e345-4d31-a006-e48048a69469\", \"fd292866-84f1-402a-9736-1d48ba26f405\": \"fd292866-84f1-402a-9736-1d48ba26f405\", \"00e8827b-1fde-4314-9937-17b05d3dcb5a\": \"00e8827b-1fde-4314-9937-17b05d3dcb5a\"}, \"doc_id_dict\": {}, \"embeddings_dict\": {}}"}}}
+{"index_store/data": {"eb3f4354-46ea-403b-b0df-7cf7482799d9": {"__type__": "vector_store", "__data__": "{\"index_id\": \"eb3f4354-46ea-403b-b0df-7cf7482799d9\", \"summary\": null, \"nodes_dict\": {\"d6b5ab1d-0961-4e99-a108-30d70f8abef2\": \"d6b5ab1d-0961-4e99-a108-30d70f8abef2\", \"c09194cd-6361-42ac-b9e7-ba087c74d7e2\": \"c09194cd-6361-42ac-b9e7-ba087c74d7e2\", \"5532cc4a-e055-410c-a373-49430e34d5d1\": \"5532cc4a-e055-410c-a373-49430e34d5d1\", \"1a3a3b7c-ee15-4d44-bae3-50a77f944c8d\": \"1a3a3b7c-ee15-4d44-bae3-50a77f944c8d\", \"f192b2db-6de7-4806-9c6c-8e1c3abc1d93\": \"f192b2db-6de7-4806-9c6c-8e1c3abc1d93\", \"56da22db-41b3-4aaf-808d-821deae2e5e4\": \"56da22db-41b3-4aaf-808d-821deae2e5e4\", \"05751854-84bb-4efb-aac7-1b9c0963712a\": \"05751854-84bb-4efb-aac7-1b9c0963712a\", \"c6a53284-e08f-4c91-9247-67afe3d6a669\": \"c6a53284-e08f-4c91-9247-67afe3d6a669\", \"3549011f-8621-461a-8d53-00df3e0ba5cb\": \"3549011f-8621-461a-8d53-00df3e0ba5cb\", \"9c6fc2b6-6251-4321-bc19-11724a6c4ce2\": \"9c6fc2b6-6251-4321-bc19-11724a6c4ce2\", \"788851b8-e719-4796-86ba-109ea1e03798\": \"788851b8-e719-4796-86ba-109ea1e03798\", \"df752e78-023d-4043-b0f9-a822ac6ebb6c\": \"df752e78-023d-4043-b0f9-a822ac6ebb6c\", \"7fb24649-a9ee-4879-95a5-576f1218764b\": \"7fb24649-a9ee-4879-95a5-576f1218764b\", \"2fcc3328-92c5-4a67-8bf3-077a0f0e0360\": \"2fcc3328-92c5-4a67-8bf3-077a0f0e0360\", \"e7923477-3d12-4cc7-9ec6-8074aa07a454\": \"e7923477-3d12-4cc7-9ec6-8074aa07a454\", \"4dc49f3e-c065-4ed5-9dff-b0a9a9f2f521\": \"4dc49f3e-c065-4ed5-9dff-b0a9a9f2f521\", \"d4e8af6d-31e3-4394-a497-198261b0be54\": \"d4e8af6d-31e3-4394-a497-198261b0be54\", \"9d3ab415-92f1-4d4c-ada2-3df199495f44\": \"9d3ab415-92f1-4d4c-ada2-3df199495f44\", \"e91808ff-f72e-4cd2-a10e-7b7027c3117e\": \"e91808ff-f72e-4cd2-a10e-7b7027c3117e\", \"88b792b9-c388-4e3b-a521-9213b14898ff\": \"88b792b9-c388-4e3b-a521-9213b14898ff\", \"79b10ad5-6204-4319-b7ff-1168bb0268ba\": \"79b10ad5-6204-4319-b7ff-1168bb0268ba\", \"2780b88e-9078-43df-a63d-9a8c6be609ba\": \"2780b88e-9078-43df-a63d-9a8c6be609ba\", \"889775ae-5120-4b1c-8b55-929e79c14bb8\": \"889775ae-5120-4b1c-8b55-929e79c14bb8\", \"6c38ece5-f833-48a8-a596-699a8eaebd4d\": \"6c38ece5-f833-48a8-a596-699a8eaebd4d\", \"160f5173-ce59-4354-bb1b-2ef61fda44bd\": \"160f5173-ce59-4354-bb1b-2ef61fda44bd\", \"b672476d-0dd7-453f-9a5b-8b198624b7e4\": \"b672476d-0dd7-453f-9a5b-8b198624b7e4\", \"46074827-aca5-4bc0-8450-1cc7ae4b8581\": \"46074827-aca5-4bc0-8450-1cc7ae4b8581\", \"da081514-32e3-43c9-a641-fc4961aeb8f1\": \"da081514-32e3-43c9-a641-fc4961aeb8f1\", \"29bf10fc-248f-4095-9e35-4f679dfebc96\": \"29bf10fc-248f-4095-9e35-4f679dfebc96\", \"d94e2a2a-53c5-4efa-946c-a6cdfb2b28aa\": \"d94e2a2a-53c5-4efa-946c-a6cdfb2b28aa\", \"510591bf-3ea8-4a9c-ae4b-80994bc8bc9f\": \"510591bf-3ea8-4a9c-ae4b-80994bc8bc9f\", \"026deff9-e172-4d98-82cc-712515a0d142\": \"026deff9-e172-4d98-82cc-712515a0d142\", \"41865b0a-f466-4af8-bb63-d65448869e93\": \"41865b0a-f466-4af8-bb63-d65448869e93\", \"8eee4a42-8c54-47d8-b806-f666e7dff495\": \"8eee4a42-8c54-47d8-b806-f666e7dff495\", \"a1d9a7e2-807c-4234-87aa-fb04d212b1d6\": \"a1d9a7e2-807c-4234-87aa-fb04d212b1d6\", \"c002f794-a118-4596-8fc5-1a583fc1cd8e\": \"c002f794-a118-4596-8fc5-1a583fc1cd8e\", \"35e2081d-5288-4b59-8e8d-3e956a47ae9c\": \"35e2081d-5288-4b59-8e8d-3e956a47ae9c\", \"ddf2f48d-33dc-4125-bbfa-d246c1cf5f9c\": \"ddf2f48d-33dc-4125-bbfa-d246c1cf5f9c\", \"d601700a-368f-4017-b334-d46d10325bbb\": \"d601700a-368f-4017-b334-d46d10325bbb\", \"fbae5f18-728a-4c9e-a69a-4dcc99522076\": \"fbae5f18-728a-4c9e-a69a-4dcc99522076\", \"5dcf1ce1-4668-401c-8c9a-6fac43a7d6a5\": \"5dcf1ce1-4668-401c-8c9a-6fac43a7d6a5\", \"bb824fc5-0534-4c58-a06a-ed71019e476d\": \"bb824fc5-0534-4c58-a06a-ed71019e476d\", \"b85898f8-4f69-4654-ae10-15ef39064e9c\": \"b85898f8-4f69-4654-ae10-15ef39064e9c\", \"71393b4c-c46e-4772-ae25-4900737e9f88\": \"71393b4c-c46e-4772-ae25-4900737e9f88\", \"87f2210c-e344-4a79-8168-3e0b8a94f5c3\": \"87f2210c-e344-4a79-8168-3e0b8a94f5c3\", \"13f2a3a8-2c0d-4b42-a9c6-a5719324f406\": \"13f2a3a8-2c0d-4b42-a9c6-a5719324f406\", \"b2d85a76-08fc-45aa-94c2-177c8996dcc1\": \"b2d85a76-08fc-45aa-94c2-177c8996dcc1\", \"ac709021-f911-4b35-be05-f3b3411f7d9c\": \"ac709021-f911-4b35-be05-f3b3411f7d9c\", \"096927c1-1464-4bbe-8d55-39811b68b384\": \"096927c1-1464-4bbe-8d55-39811b68b384\", \"c3d6c75e-a3ef-443a-b724-199be3dc9897\": \"c3d6c75e-a3ef-443a-b724-199be3dc9897\", \"f082491a-5cfc-49c0-9c75-1b8f8cbe283f\": \"f082491a-5cfc-49c0-9c75-1b8f8cbe283f\", \"d7bffc3e-0af8-49ec-a00c-75692643ffa1\": \"d7bffc3e-0af8-49ec-a00c-75692643ffa1\"}, \"doc_id_dict\": {}, \"embeddings_dict\": {}}"}}}