update config

chatgpt_long_term_memory/__init__.py

@@ -1,3 +1,6 @@
+from chatgpt_long_term_memory.openai_engine import (OpenAIChatBot,
+                                                  OpenAIChatConfig,
+                                                  retry_on_openai_errors)
 from chatgpt_long_term_memory.llama_index_helpers import DocIndexer, Retrievers
 from chatgpt_long_term_memory.llama_index_helpers.config import (
     IndexConfig, RetrieversConfig)
@@ -10,5 +13,8 @@
     "IndexConfig",
     "RetrieversConfig",
     "ChatMemory",
-    "ChatMemoryConfig"
+    "ChatMemoryConfig",
+    "OpenAIChatConfig",
+    "OpenAIChatBot",
+    "retry_on_openai_errors"
 ]

chatgpt_long_term_memory/llama_index_helpers/config.py

@@ -1,17 +1,17 @@
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
 
 
 class IndexConfig(BaseModel):
-    root_path: str = ""
-    knowledge_base: bool = True
-    model_name: str = "gpt-3.5-turbo"
-    temperature: int = 0
-    context_window: int = 4096
-    num_outputs: int = 700
-    max_chunk_overlap: float = 0.5
-    chunk_size_limit: int = 600
+    root_path: str = Field(default="/home/elmira/projects/0-NAVIS/long-term-memory-chatbot")
+    knowledge_base: bool = Field(default=True)
+    model_name: str = Field(default="gpt-3.5-turbo")
+    temperature: int = Field(default=0)
+    context_window: int = Field(default=4096)
+    num_outputs: int = Field(default=700)
+    max_chunk_overlap: float = Field(default=0.5)
+    chunk_size_limit: int = Field(default=600)
 
 
 class RetrieversConfig(BaseModel):
-    top_k: int = 3
-    similarity_threshold: float = 0.7
+    top_k: int = Field(default=7)
+    similarity_threshold: float = Field(default=0.7)

chatgpt_long_term_memory/llama_index_helpers/index_engine.py

@@ -34,26 +34,28 @@ class DocIndexer:
     """
 
     def __init__(
-            self, config: IndexConfig):
-        self.knowledge_base = config.knowledge_base
+            self, doc_config: IndexConfig, **kw):
+        super().__init__(**kw)
+        self.config = doc_config
 
         # Initialize the OpenAI language model
-        self.llm = OpenAI(model=config.model_name,
-                          temperature=config.temperature)
+        self.llm = OpenAI(model=self.config.model_name,
+                          temperature=self.config.temperature)
         service_context = ServiceContext.from_defaults(llm=self.llm)
         set_global_service_context(service_context)
 
-        self.root_path = config.root_path
-        self.data_path = f'{self.root_path}/resources/data'
-        assert os.path.exists(
-            self.data_path), f"Path '{self.data_path}' does not exist!"
+        self.root_path = self.config.root_path
+        if self.config.knowledge_base:
+            self.data_path = f'{self.root_path}/resources/data'
+            assert os.path.exists(
+                self.data_path), f"Path '{self.data_path}' does not exist!"
 
         # Define prompt helper for the index
         self.prompt_helper = PromptHelper(
-            config.context_window,
-            config.num_outputs,
-            config.max_chunk_overlap,
-            chunk_size_limit=config.chunk_size_limit
+            self.config.context_window,
+            self.config.num_outputs,
+            self.config.max_chunk_overlap,
+            chunk_size_limit=self.config.chunk_size_limit
         )
 
     def load_documents(self, retrieved_documents):
@@ -70,7 +72,7 @@ def load_documents(self, retrieved_documents):
         sorted_data = sorted(retrieved_documents,
                              key=lambda x: list(x.keys())[0], reverse=True)
         doc = list(sorted_data[0].values())[0]
-        doc = f"question: {doc['user_query']}, answer: {doc['bot_response']}"
+        doc = f"USER: {doc['user_query']}, ANSWER: {doc['bot_response']}"
         return Document(text=doc, doc_id=f"doc_id_{str(uuid.uuid4())}")
 
     def construct_index_general(self, user_id, path, mode):
@@ -86,15 +88,18 @@ def construct_index_general(self, user_id, path, mode):
             VectorStoreIndex: The constructed index.
         """
         # If the user's storage directory does not exist, create it.
+        print(self.config.knowledge_base)
+        print(path)
         if not os.path.exists(path):
             os.makedirs(path)
 
         # Load data from the directory for the user's personal knowledge base
         if mode == 'kb':
-            if self.knowledge_base:
+            if self.config.knowledge_base:
                 documents = SimpleDirectoryReader(self.data_path).load_data()
             else:
                 # User doesn't have any personal knowledge base
+                print("kb f")
                 documents = []
 
         # Create a Document structure for the user's input query and chatbot response
@@ -125,9 +130,11 @@ def load_index(self, user_id):
         index_path = f'{self.root_path}/storages/storage_{user_id}'
         status = os.path.exists(f'{index_path}/vector_store.json')
         if status:
+            print("index_path: if ", index_path)
             index = load_index_from_storage(
                 StorageContext.from_defaults(persist_dir=index_path))
         else:
+            print("index_path: else", index_path)
             index = self.construct_index_general(
                 user_id, index_path, mode="kb")
         return index

chatgpt_long_term_memory/llama_index_helpers/retrievers_engine.py

@@ -25,9 +25,11 @@ class Retrievers:
 
     """
 
-    def __init__(self, config: RetrieversConfig):
-        self.top_k = config.top_k
-        self.similarity_threshold = config.similarity_threshold
+    def __init__(self, retrieve_config: RetrieversConfig, **kw):
+        super().__init__(**kw)
+        self.config = retrieve_config
+        self.top_k = self.config.top_k
+        self.similarity_threshold = self.config.similarity_threshold
 
     def query(self, index, question):
         """

chatgpt_long_term_memory/memory/chat_memory.py

@@ -15,9 +15,11 @@ class ChatMemory:
 
     """
 
-    def __init__(self, config: ChatMemoryConfig):
+    def __init__(self, memory_config: ChatMemoryConfig, **kw):
+        super().__init__(**kw)
+        self.config = memory_config
         self.redis_db = RedisManager(
-            host=config.redis_host, port=config.redis_port)
+            host=self.config.redis_host, port=self.config.redis_port)
 
     def get(self, user_id):
         """

chatgpt_long_term_memory/memory/config.py

@@ -1,6 +1,6 @@
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
 
 
 class ChatMemoryConfig(BaseModel):
-    redis_host: str = "172.16.0.2"
-    redis_port: int = 6379
+    redis_host: str = Field(default="172.16.0.2")
+    redis_port: int = Field(default=6379)