switch to load from vector store from qdrant cloud

app.py

@@ -20,16 +20,21 @@ from llama_index import set_global_service_context
 
 from service_provider_config import get_service_provider_config
 
-
+load_dotenv()
 # initial service setup
 px.launch_app()
 llama_index.set_global_handler("arize_phoenix")
-
-load_dotenv()
+# llama_index.set_global_handler("wandb", run_args={"project": "llamaindex"})
 openai.api_key = os.getenv("OPENAI_API_KEY")
+
+IS_LOAD_FROM_VECTOR_STORE = True 
+VDB_COLLECTION_NAME = "demo-v0"
+MODEL_NAME = ChatbotVersion.CHATGPT_35.value
+
+
 CHUNK_SIZE = 1024
 LLM, EMBED_MODEL = get_service_provider_config(
-    service_provider=ServiceProvider.OPENAI)
+    service_provider=ServiceProvider.OPENAI, model_name=MODEL_NAME)
 service_context = ServiceContext.from_defaults(
     chunk_size=CHUNK_SIZE,
     llm=LLM,
@@ -69,15 +74,15 @@ class AwesumIndexBuilder(IndexBuilder):
         pipeline = IngestionPipeline(
             transformations=[
                 SentenceSplitter(),
-                EMBED_MODEL,
+                self.embed_model,
             ],
             vector_store=self.vector_store,
         )
-        pipeline.run(documents=self.documents)
+        pipeline.run(documents=self.documents, show_progress=True)
         self.index = VectorStoreIndex.from_vector_store(self.vector_store)
 
 
-class AwesumCareChatbot(Chatbot):
+class AwesumCareToolChatbot(Chatbot):
     DENIED_ANSWER_PROMPT = ""
     SYSTEM_PROMPT = ""
     CHAT_EXAMPLES = [
@@ -126,13 +131,48 @@ class AwesumCareChatbot(Chatbot):
         #     text_qa_template=CHAT_TEXT_QA_PROMPT)
         super()._setup_chat_engine()
 
+class AweSumCareContextChatbot(AwesumCareToolChatbot):
+    def _setup_query_engine(self):
+        pass
+    def _setup_tools(self):
+        pass
+    def _setup_chat_engine(self):
+        self.chat_engine = self.index.as_chat_engine(
+            chat_mode=ChatMode.CONTEXT,
+            similarity_top_k=5,
+            text_qa_template=CHAT_TEXT_QA_PROMPT)
+
+class AweSumCareSimpleChatbot(AwesumCareToolChatbot):
+    def _setup_query_engine(self):
+        pass
+    def _setup_tools(self):
+        pass
+    def _setup_chat_engine(self):
+        self.chat_engine = self.index.as_chat_engine(
+            chat_mode=ChatMode.SIMPLE)
+
+model_name = MODEL_NAME
+index_builder = AwesumIndexBuilder(vdb_collection_name=VDB_COLLECTION_NAME,
+                                   embed_model=EMBED_MODEL,
+                                   is_load_from_vector_store=IS_LOAD_FROM_VECTOR_STORE)
 
 # gpt-3.5-turbo-1106, gpt-4-1106-preview
-awesum_chatbot = AwesumCareChatbot(model_name=ChatbotVersion.CHATGPT_35.value,
-                                   index_builder=AwesumIndexBuilder(
-                                       vdb_collection_name="demo-v0",
-                                       is_load_from_vector_store=False)
-                                   )
+awesum_chatbot = AwesumCareToolChatbot(model_name=model_name, index_builder=index_builder)
+awesum_chatbot_context = AweSumCareContextChatbot(model_name=model_name, index_builder=index_builder)
+awesum_chatbot_simple = AweSumCareSimpleChatbot(model_name=model_name, index_builder=index_builder)
+
+
+def service_setup(model_name):
+    CHUNK_SIZE = 1024
+    LLM, EMBED_MODEL = get_service_provider_config(
+        service_provider=ServiceProvider.OPENAI, model_name=model_name)
+    service_context = ServiceContext.from_defaults(
+        chunk_size=CHUNK_SIZE,
+        llm=LLM,
+        embed_model=EMBED_MODEL,
+    )
+    set_global_service_context(service_context)
+    return LLM, EMBED_MODEL
 
 
 def vote(data: gr.LikeData):
@@ -145,23 +185,62 @@ def vote(data: gr.LikeData):
 chatbot = gr.Chatbot()
 
 with gr.Blocks() as demo:
+
     gr.Markdown("# Awesum Care demo")
 
-    with gr.Tab("With awesum care data prepared"):
-        gr.ChatInterface(
+    # with gr.Row():
+    #     model_selector = gr.Radio(
+    #         value=ChatbotVersion.CHATGPT_35.value,
+    #         choices=[ChatbotVersion.CHATGPT_35.value, ChatbotVersion.CHATGPT_4.value],
+    #         label="Select Chatbot Model (To be implemented)"
+    #     )
+
+    with gr.Tab("With context aware"):
+        context_interface = gr.ChatInterface(
+            awesum_chatbot_context.stream_chat,
+            examples=awesum_chatbot.CHAT_EXAMPLES,
+        )
+        chatbot.like(vote, None, None)
+
+    with gr.Tab("With function calling as tool to retrieve"):
+        function_call_interface = gr.ChatInterface(
             awesum_chatbot.stream_chat,
-            chatbot=chatbot,
             examples=awesum_chatbot.CHAT_EXAMPLES,
         )
         chatbot.like(vote, None, None)
 
-    with gr.Tab("With Initial System Prompt (a.k.a. prompt wrapper)"):
-        gr.ChatInterface(
-            awesum_chatbot.predict_with_prompt_wrapper, examples=awesum_chatbot.CHAT_EXAMPLES)
 
     with gr.Tab("Vanilla ChatGPT without modification"):
-        gr.ChatInterface(awesum_chatbot.predict_vanilla_chatgpt,
-                         examples=awesum_chatbot.CHAT_EXAMPLES)
+        vanilla_interface = gr.ChatInterface(
+            awesum_chatbot_simple.stream_chat,
+            examples=awesum_chatbot.CHAT_EXAMPLES)
+
+    # @model_selector.change(inputs=[model_selector, chatbot], outputs=[context_interface, function_call_interface, vanilla_interface])
+    # def switch_model(model_name, my_chatbot):
+    #     print(model_name)
+    #     print(my_chatbot.config())
+    #     LLM, EMBED_MODEL = service_setup(model_name)
+    #     # global awesum_chatbot, awesum_chatbot_context, awesum_chatbot_simple
+    #     # Logic to switch models - create new instances of the chatbots with the selected model
+    #     index_builder = AwesumIndexBuilder(vdb_collection_name=VDB_COLLECTION_NAME,
+    #                                     embed_model=EMBED_MODEL,
+    #                                     is_load_from_vector_store=IS_LOAD_FROM_VECTOR_STORE)
+    #     awesum_chatbot = AwesumCareToolChatbot(model_name=model_name, index_builder=index_builder, llm=LLM)
+    #     awesum_chatbot_context = AweSumCareContextChatbot(model_name=model_name, index_builder=index_builder)
+    #     awesum_chatbot_simple = AweSumCareSimpleChatbot(model_name=model_name, index_builder=index_builder)
+    #     # return awesum_chatbot.stream_chat, awesum_chatbot_context.stream_chat, awesum_chatbot_simple.stream_chat
+
+    #     new_context_interface = gr.ChatInterface(
+    #         awesum_chatbot_context.stream_chat,
+    #     )
+    #     new_function_call_interface = gr.ChatInterface(
+    #         awesum_chatbot.stream_chat,
+    #     )
+    #     new_vanilla_interface = gr.ChatInterface(
+    #         awesum_chatbot_simple.stream_chat,
+    #     )
+    #     return new_context_interface, new_function_call_interface, new_vanilla_interface
+
 
 demo.queue()
-demo.launch()
+demo.launch(share=False)

chatbot.py

@@ -1,4 +1,5 @@
 from enum import Enum
+import logging
 from typing import List
 import os
 import re
@@ -8,15 +9,16 @@ from openai import OpenAI
 import phoenix as px
 import llama_index
 from llama_index import OpenAIEmbedding
-from llama_index.llms import ChatMessage, MessageRole, OpenAI
+from llama_index.llms import ChatMessage, MessageRole
 
 load_dotenv()
 
 
 class IndexBuilder:
-    def __init__(self, vdb_collection_name, is_load_from_vector_store=False):
+    def __init__(self, vdb_collection_name, embed_model, is_load_from_vector_store=False):
         self.documents = None
         self.vdb_collection_name = vdb_collection_name
+        self.embed_model = embed_model 
         self.index = None
         self.is_load_from_vector_store = is_load_from_vector_store
         self.build_index()
@@ -52,9 +54,10 @@ class Chatbot:
     DENIED_ANSWER_PROMPT = ""
     CHAT_EXAMPLES = []
 
-    def __init__(self, model_name, index_builder: IndexBuilder):
+    def __init__(self, model_name, index_builder: IndexBuilder, llm=None):
         self.model_name = model_name
         self.index_builder = index_builder
+        self.llm = llm
 
         self.documents = None
         self.index = None
@@ -63,8 +66,23 @@ class Chatbot:
         self.vector_store = None
         self.tools = None
 
+        self._setup_logger()
         self._setup_chatbot()
 
+
+    def _setup_logger(self):
+        logs_dir = 'logs'
+        if not os.path.exists(logs_dir):
+            os.makedirs(logs_dir)  # Step 3: Create logs directory
+
+        logging.basicConfig(
+            filename=os.path.join(logs_dir, 'chatbot.log'),
+            filemode='a',
+            format='%(asctime)s - %(levelname)s - %(message)s',
+            level=logging.INFO
+        )
+        self.logger = logging.getLogger(__name__)
+
     def _setup_chatbot(self):
         # self._setup_observer()
         self._setup_index()
@@ -97,8 +115,8 @@ class Chatbot:
         print("Setup chat engine...")
 
     def stream_chat(self, message, history):
-        print(history)
-        print(self.convert_to_chat_messages(history))
+        self.logger.info(history)
+        self.logger.info(self.convert_to_chat_messages(history))
         response = self.chat_engine.stream_chat(
             message, chat_history=self.convert_to_chat_messages(history)
         )
@@ -145,15 +163,17 @@ class Chatbot:
                 {"role": "assistant", "content": assistant})
         history_openai_format.append({"role": "user", "content": message})
 
-        import openai
-        print(openai.__version__)
         stream = openai_client.chat.completions.create(
             model=self.model_name,
             messages=history_openai_format,
             temperature=1.0,
             stream=True)
+
+        partial_message = ""
         for part in stream:
-            yield part.choices[0].delta.content or ""
+            partial_message += part.choices[0].delta.content or ""
+            yield partial_message
+            # yield part.choices[0].delta.content or ""
         # partial_message = ""
         # for chunk in response:
         #     if len(chunk["choices"][0]["delta"]) != 0:

qdrant.py

@@ -1,5 +1,10 @@
+# import qdrant_client
+from qdrant_client import QdrantClient
 
-import qdrant_client
 
+# client = qdrant_client.QdrantClient(path="/tmp/total_qdrant/")
 
-client = qdrant_client.QdrantClient(path="/tmp/total_qdrant/")
+client = QdrantClient(
+    url="https://dc059edb-e5bc-43ff-bb72-756cc610d6d1.us-east4-0.gcp.cloud.qdrant.io",
+    api_key="th0__GAZkpGRP0QSL2QwB0g_vr0ATvY-sd2Gre5VBUk8-vExdsYsYQ"
+)

service_provider_config.py

@@ -7,11 +7,11 @@ from schemas import ServiceProvider, ChatbotVersion
 
 load_dotenv()
 
-def get_service_provider_config(service_provider: ServiceProvider):
+def get_service_provider_config(service_provider: ServiceProvider, model_name: str=ChatbotVersion.CHATGPT_35.value):
     if service_provider == ServiceProvider.AZURE:
         return get_azure_openai_config()
     if service_provider == ServiceProvider.OPENAI:
-        llm = OpenAI(model=ChatbotVersion.CHATGPT_35)
+        llm = OpenAI(model=model_name)
         embed_model = OpenAIEmbedding()
         return llm, embed_model