diff --git a/examples/lightrag_api_llamaindex_direct_demo_simplified.py b/examples/lightrag_llamaindex_direct_demo.py
similarity index 95%
rename from examples/lightrag_api_llamaindex_direct_demo_simplified.py
rename to examples/lightrag_llamaindex_direct_demo.py
index a1781842..5db158ce 100644
--- a/examples/lightrag_api_llamaindex_direct_demo_simplified.py
+++ b/examples/lightrag_llamaindex_direct_demo.py
@@ -1,6 +1,6 @@
 import os
 from lightrag import LightRAG, QueryParam
-from lightrag.wrapper.llama_index_impl import (
+from lightrag.llm.llama_index_impl import (
     llama_index_complete_if_cache,
     llama_index_embed,
 )
@@ -10,14 +10,13 @@ from llama_index.embeddings.openai import OpenAIEmbedding
 import asyncio
 
 # Configure working directory
-DEFAULT_RAG_DIR = "index_default"
-WORKING_DIR = os.environ.get("RAG_DIR", f"{DEFAULT_RAG_DIR}")
+WORKING_DIR = "./index_default"
 print(f"WORKING_DIR: {WORKING_DIR}")
 
 # Model configuration
 LLM_MODEL = os.environ.get("LLM_MODEL", "gpt-4")
 print(f"LLM_MODEL: {LLM_MODEL}")
-EMBEDDING_MODEL = os.environ.get("EMBEDDING_MODEL", "text-embedding-3-small")
+EMBEDDING_MODEL = os.environ.get("EMBEDDING_MODEL", "text-embedding-3-large")
 print(f"EMBEDDING_MODEL: {EMBEDDING_MODEL}")
 EMBEDDING_MAX_TOKEN_SIZE = int(os.environ.get("EMBEDDING_MAX_TOKEN_SIZE", 8192))
 print(f"EMBEDDING_MAX_TOKEN_SIZE: {EMBEDDING_MAX_TOKEN_SIZE}")
@@ -26,6 +25,7 @@ print(f"EMBEDDING_MAX_TOKEN_SIZE: {EMBEDDING_MAX_TOKEN_SIZE}")
 OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY", "your-api-key-here")
 
 if not os.path.exists(WORKING_DIR):
+    print(f"Creating working directory: {WORKING_DIR}")
     os.mkdir(WORKING_DIR)
 
 
diff --git a/examples/lightrag_api_llamaindex_litellm_demo_simplified.py b/examples/lightrag_llamaindex_litellm_demo.py
similarity index 92%
rename from examples/lightrag_api_llamaindex_litellm_demo_simplified.py
rename to examples/lightrag_llamaindex_litellm_demo.py
index a1ab90db..3511ecf3 100644
--- a/examples/lightrag_api_llamaindex_litellm_demo_simplified.py
+++ b/examples/lightrag_llamaindex_litellm_demo.py
@@ -1,6 +1,6 @@
 import os
 from lightrag import LightRAG, QueryParam
-from lightrag.wrapper.llama_index_impl import (
+from lightrag.llm.llama_index_impl import (
     llama_index_complete_if_cache,
     llama_index_embed,
 )
@@ -10,14 +10,13 @@ from llama_index.embeddings.litellm import LiteLLMEmbedding
 import asyncio
 
 # Configure working directory
-DEFAULT_RAG_DIR = "index_default"
-WORKING_DIR = os.environ.get("RAG_DIR", f"{DEFAULT_RAG_DIR}")
+WORKING_DIR = "./index_default"
 print(f"WORKING_DIR: {WORKING_DIR}")
 
 # Model configuration
-LLM_MODEL = os.environ.get("LLM_MODEL", "gpt-4o")
+LLM_MODEL = os.environ.get("LLM_MODEL", "gpt-4")
 print(f"LLM_MODEL: {LLM_MODEL}")
-EMBEDDING_MODEL = os.environ.get("EMBEDDING_MODEL", "embedding-model")
+EMBEDDING_MODEL = os.environ.get("EMBEDDING_MODEL", "text-embedding-3-large")
 print(f"EMBEDDING_MODEL: {EMBEDDING_MODEL}")
 EMBEDDING_MAX_TOKEN_SIZE = int(os.environ.get("EMBEDDING_MAX_TOKEN_SIZE", 8192))
 print(f"EMBEDDING_MAX_TOKEN_SIZE: {EMBEDDING_MAX_TOKEN_SIZE}")
diff --git a/lightrag/wrapper/Readme.md b/lightrag/llm/Readme.md
similarity index 100%
rename from lightrag/wrapper/Readme.md
rename to lightrag/llm/Readme.md
diff --git a/lightrag/wrapper/llama_index_impl.py b/lightrag/llm/llama_index_impl.py
similarity index 100%
rename from lightrag/wrapper/llama_index_impl.py
rename to lightrag/llm/llama_index_impl.py
diff --git a/lightrag/wrapper/__init__.py b/lightrag/wrapper/__init__.py
deleted file mode 100644
index e69de29b..00000000