Merge remote-tracking branch 'origin/main'

LICENSE

@@ -1,6 +1,6 @@
 MIT License
 
-Copyright (c) 2025 Gustavo Ye
+Copyright (c) 2025 LarFii
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

README.md

@@ -12,7 +12,7 @@
     </p>
      <p>
           <img src='https://img.shields.io/github/stars/hkuds/lightrag?color=green&style=social' />
-        <img src="https://img.shields.io/badge/python->=3.10-blue">
+        <img src="https://img.shields.io/badge/python-3.10-blue">
         <a href="https://pypi.org/project/lightrag-hku/"><img src="https://img.shields.io/pypi/v/lightrag-hku.svg"></a>
         <a href="https://pepy.tech/project/lightrag-hku"><img src="https://static.pepy.tech/badge/lightrag-hku/month"></a>
     </p>
@@ -637,7 +637,7 @@ if __name__ == "__main__":
 | **llm\_model\_kwargs**                       | `dict` | Additional parameters for LLM generation                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            |                                                                                                             |
 | **vector\_db\_storage\_cls\_kwargs**         | `dict` | Additional parameters for vector database (currently not used)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                      |                                                                                                             |
 | **enable\_llm\_cache**                       | `bool` | If `TRUE`, stores LLM results in cache; repeated prompts return cached responses                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    | `TRUE`                                                                                                      |
-| **enable\_llm\_cache\_for\_entity\_extract** | `bool` | If `TRUE`, stores LLM results in cache for entity extraction; Good for beginners to debug your application                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                  | `FALSE`                                                                                                     |
+| **enable\_llm\_cache\_for\_entity\_extract** | `bool` | If `TRUE`, stores LLM results in cache for entity extraction; Good for beginners to debug your application                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                  | `TRUE`                                                                                                     |
 | **addon\_params**                            | `dict` | Additional parameters, e.g., `{"example_number": 1, "language": "Simplified Chinese", "entity_types": ["organization", "person", "geo", "event"], "insert_batch_size": 10}`: sets example limit, output language, and batch size for document processing                                                                                                                                                                                                                                                                                                                                                                                                                            | `example_number: all examples, language: English, insert_batch_size: 10`                                    |
 | **convert\_response\_to\_json\_func**        | `callable` | Not used                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            | `convert_response_to_json`                                                                                  |
 | **embedding\_cache\_config**                 | `dict` | Configuration for question-answer caching. Contains three parameters:<br>- `enabled`: Boolean value to enable/disable cache lookup functionality. When enabled, the system will check cached responses before generating new answers.<br>- `similarity_threshold`: Float value (0-1), similarity threshold. When a new question's similarity with a cached question exceeds this threshold, the cached answer will be returned directly without calling the LLM.<br>- `use_llm_check`: Boolean value to enable/disable LLM similarity verification. When enabled, LLM will be used as a secondary check to verify the similarity between questions before returning cached answers. | Default: `{"enabled": False, "similarity_threshold": 0.95, "use_llm_check": False}`                         |
@@ -892,69 +892,6 @@ def extract_queries(file_path):
 ```
 </details>
 
-## Code Structure
-
-```python
-.
-├── .github/
-│   ├── workflows/
-│   │   └── linting.yaml
-├── examples/
-│   ├── batch_eval.py
-│   ├── generate_query.py
-│   ├── graph_visual_with_html.py
-│   ├── graph_visual_with_neo4j.py
-│   ├── insert_custom_kg.py
-│   ├── lightrag_api_openai_compatible_demo.py
-│   ├── lightrag_api_oracle_demo..py
-│   ├── lightrag_azure_openai_demo.py
-│   ├── lightrag_bedrock_demo.py
-│   ├── lightrag_hf_demo.py
-│   ├── lightrag_lmdeploy_demo.py
-│   ├── lightrag_nvidia_demo.py
-│   ├── lightrag_ollama_demo.py
-│   ├── lightrag_openai_compatible_demo.py
-│   ├── lightrag_openai_demo.py
-│   ├── lightrag_oracle_demo.py
-│   ├── lightrag_siliconcloud_demo.py
-│   └── vram_management_demo.py
-├── lightrag/
-│   ├── api/
-│   │   ├── lollms_lightrag_server.py
-│   │   ├── ollama_lightrag_server.py
-│   │   ├── openai_lightrag_server.py
-│   │   ├── azure_openai_lightrag_server.py
-│   │   └── requirements.txt
-│   ├── kg/
-│   │   ├── __init__.py
-│   │   ├── oracle_impl.py
-│   │   └── neo4j_impl.py
-│   ├── __init__.py
-│   ├── base.py
-│   ├── lightrag.py
-│   ├── llm.py
-│   ├── operate.py
-│   ├── prompt.py
-│   ├── storage.py
-│   └── utils.py
-├── reproduce/
-│   ├── Step_0.py
-│   ├── Step_1_openai_compatible.py
-│   ├── Step_1.py
-│   ├── Step_2.py
-│   ├── Step_3_openai_compatible.py
-│   └── Step_3.py
-├── .gitignore
-├── .pre-commit-config.yaml
-├── get_all_edges_nx.py
-├── LICENSE
-├── README.md
-├── requirements.txt
-├── setup.py
-├── test_neo4j.py
-└── test.py
-```
-
 ## Install with API Support
 
 LightRAG provides optional API support through FastAPI servers that add RAG capabilities to existing LLM services. You can install LightRAG with API support in two ways:

examples/test_split_by_character.ipynb

@@ -0,0 +1,1296 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "4b5690db12e34685",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-01-09T03:40:58.307102Z",
+     "start_time": "2025-01-09T03:40:51.935233Z"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "import os\n",
+    "import logging\n",
+    "import numpy as np\n",
+    "from lightrag import LightRAG, QueryParam\n",
+    "from lightrag.llm import openai_complete_if_cache, openai_embedding\n",
+    "from lightrag.utils import EmbeddingFunc\n",
+    "import nest_asyncio"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "dd17956ec322b361",
+   "metadata": {},
+   "source": "#### split by character"
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "8c8ee7c061bf9159",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-01-09T03:41:13.961167Z",
+     "start_time": "2025-01-09T03:41:13.958357Z"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "nest_asyncio.apply()\n",
+    "WORKING_DIR = \"../../llm_rag/paper_db/R000088_test1\"\n",
+    "logging.basicConfig(format=\"%(levelname)s:%(message)s\", level=logging.INFO)\n",
+    "if not os.path.exists(WORKING_DIR):\n",
+    "    os.mkdir(WORKING_DIR)\n",
+    "API = os.environ.get(\"DOUBAO_API_KEY\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "a5009d16e0851dca",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-01-09T03:41:16.862036Z",
+     "start_time": "2025-01-09T03:41:16.859306Z"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "async def llm_model_func(\n",
+    "    prompt, system_prompt=None, history_messages=[], keyword_extraction=False, **kwargs\n",
+    ") -> str:\n",
+    "    return await openai_complete_if_cache(\n",
+    "        \"ep-20241218114828-2tlww\",\n",
+    "        prompt,\n",
+    "        system_prompt=system_prompt,\n",
+    "        history_messages=history_messages,\n",
+    "        api_key=API,\n",
+    "        base_url=\"https://ark.cn-beijing.volces.com/api/v3\",\n",
+    "        **kwargs,\n",
+    "    )\n",
+    "\n",
+    "\n",
+    "async def embedding_func(texts: list[str]) -> np.ndarray:\n",
+    "    return await openai_embedding(\n",
+    "        texts,\n",
+    "        model=\"ep-20241231173413-pgjmk\",\n",
+    "        api_key=API,\n",
+    "        base_url=\"https://ark.cn-beijing.volces.com/api/v3\",\n",
+    "    )"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "397fcad24ce4d0ed",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-01-09T03:41:24.950307Z",
+     "start_time": "2025-01-09T03:41:24.940353Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "INFO:lightrag:Logger initialized for working directory: ../../llm_rag/paper_db/R000088_test1\n",
+      "INFO:lightrag:Load KV llm_response_cache with 0 data\n",
+      "INFO:lightrag:Load KV full_docs with 0 data\n",
+      "INFO:lightrag:Load KV text_chunks with 0 data\n",
+      "INFO:nano-vectordb:Init {'embedding_dim': 4096, 'metric': 'cosine', 'storage_file': '../../llm_rag/paper_db/R000088_test1/vdb_entities.json'} 0 data\n",
+      "INFO:nano-vectordb:Init {'embedding_dim': 4096, 'metric': 'cosine', 'storage_file': '../../llm_rag/paper_db/R000088_test1/vdb_relationships.json'} 0 data\n",
+      "INFO:nano-vectordb:Init {'embedding_dim': 4096, 'metric': 'cosine', 'storage_file': '../../llm_rag/paper_db/R000088_test1/vdb_chunks.json'} 0 data\n",
+      "INFO:lightrag:Loaded document status storage with 0 records\n"
+     ]
+    }
+   ],
+   "source": [
+    "rag = LightRAG(\n",
+    "    working_dir=WORKING_DIR,\n",
+    "    llm_model_func=llm_model_func,\n",
+    "    embedding_func=EmbeddingFunc(\n",
+    "        embedding_dim=4096, max_token_size=8192, func=embedding_func\n",
+    "    ),\n",
+    "    chunk_token_size=512,\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "1dc3603677f7484d",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-01-09T03:41:37.947456Z",
+     "start_time": "2025-01-09T03:41:37.941901Z"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "with open(\n",
+    "    \"../../llm_rag/example/R000088/auto/R000088_full_txt.md\", \"r\", encoding=\"utf-8\"\n",
+    ") as f:\n",
+    "    content = f.read()\n",
+    "\n",
+    "\n",
+    "async def embedding_func(texts: list[str]) -> np.ndarray:\n",
+    "    return await openai_embedding(\n",
+    "        texts,\n",
+    "        model=\"ep-20241231173413-pgjmk\",\n",
+    "        api_key=API,\n",
+    "        base_url=\"https://ark.cn-beijing.volces.com/api/v3\",\n",
+    "    )\n",
+    "\n",
+    "\n",
+    "async def get_embedding_dim():\n",
+    "    test_text = [\"This is a test sentence.\"]\n",
+    "    embedding = await embedding_func(test_text)\n",
+    "    embedding_dim = embedding.shape[1]\n",
+    "    return embedding_dim"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "6844202606acfbe5",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-01-09T03:41:39.608541Z",
+     "start_time": "2025-01-09T03:41:39.165057Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n"
+     ]
+    }
+   ],
+   "source": [
+    "embedding_dimension = await get_embedding_dim()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "id": "d6273839d9681403",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-01-09T03:44:34.295345Z",
+     "start_time": "2025-01-09T03:41:48.324171Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "INFO:lightrag:Processing 1 new unique documents\n",
+      "Processing batch 1:   0%|          | 0/1 [00:00<?, ?it/s]INFO:lightrag:Inserting 35 vectors to chunks\n",
+      "\n",
+      "Generating embeddings:   0%|          | 0/2 [00:00<?, ?batch/s]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:  50%|█████     | 1/2 [00:00<00:00,  1.36batch/s]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings: 100%|██████████| 2/2 [00:04<00:00,  2.25s/batch]\u001b[A\n",
+      "\n",
+      "Extracting entities from chunks:   0%|          | 0/35 [00:00<?, ?chunk/s]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠙ Processed 1 chunks, 1 entities(duplicated), 0 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:   3%|▎         | 1/35 [00:04<02:47,  4.93s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠹ Processed 2 chunks, 2 entities(duplicated), 0 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:   6%|▌         | 2/35 [00:05<01:18,  2.37s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠸ Processed 3 chunks, 9 entities(duplicated), 5 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:   9%|▊         | 3/35 [00:26<05:43, 10.73s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠼ Processed 4 chunks, 16 entities(duplicated), 11 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  11%|█▏        | 4/35 [00:26<03:24,  6.60s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠴ Processed 5 chunks, 24 entities(duplicated), 18 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  14%|█▍        | 5/35 [00:33<03:24,  6.82s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠦ Processed 6 chunks, 35 entities(duplicated), 28 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  17%|█▋        | 6/35 [00:42<03:38,  7.53s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠧ Processed 7 chunks, 47 entities(duplicated), 36 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  20%|██        | 7/35 [00:43<02:28,  5.31s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠇ Processed 8 chunks, 61 entities(duplicated), 49 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  23%|██▎       | 8/35 [00:45<01:52,  4.16s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠏ Processed 9 chunks, 81 entities(duplicated), 49 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠋ Processed 10 chunks, 90 entities(duplicated), 62 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  29%|██▊       | 10/35 [00:46<01:06,  2.64s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠙ Processed 11 chunks, 101 entities(duplicated), 80 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  31%|███▏      | 11/35 [00:52<01:19,  3.31s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠹ Processed 12 chunks, 108 entities(duplicated), 85 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  34%|███▍      | 12/35 [00:54<01:11,  3.12s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠸ Processed 13 chunks, 120 entities(duplicated), 100 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  37%|███▋      | 13/35 [00:59<01:18,  3.55s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠼ Processed 14 chunks, 131 entities(duplicated), 110 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  40%|████      | 14/35 [01:00<00:59,  2.82s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠴ Processed 15 chunks, 143 entities(duplicated), 110 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  43%|████▎     | 15/35 [01:02<00:52,  2.64s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠦ Processed 16 chunks, 162 entities(duplicated), 124 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  46%|████▌     | 16/35 [01:05<00:53,  2.80s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠧ Processed 17 chunks, 174 entities(duplicated), 132 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  49%|████▊     | 17/35 [01:06<00:39,  2.19s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠇ Processed 18 chunks, 185 entities(duplicated), 137 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  51%|█████▏    | 18/35 [01:12<00:53,  3.15s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠏ Processed 19 chunks, 193 entities(duplicated), 149 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  54%|█████▍    | 19/35 [01:18<01:06,  4.14s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠋ Processed 20 chunks, 205 entities(duplicated), 158 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  57%|█████▋    | 20/35 [01:19<00:50,  3.35s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠙ Processed 21 chunks, 220 entities(duplicated), 187 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  60%|██████    | 21/35 [01:27<01:02,  4.47s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠹ Processed 22 chunks, 247 entities(duplicated), 216 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  63%|██████▎   | 22/35 [01:30<00:54,  4.16s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠸ Processed 23 chunks, 260 entities(duplicated), 230 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  66%|██████▌   | 23/35 [01:34<00:48,  4.05s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠼ Processed 24 chunks, 291 entities(duplicated), 253 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  69%|██████▊   | 24/35 [01:38<00:44,  4.03s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠴ Processed 25 chunks, 304 entities(duplicated), 262 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  71%|███████▏  | 25/35 [01:41<00:36,  3.67s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠦ Processed 26 chunks, 313 entities(duplicated), 271 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  74%|███████▍  | 26/35 [01:41<00:24,  2.76s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠧ Processed 27 chunks, 321 entities(duplicated), 283 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  77%|███████▋  | 27/35 [01:47<00:28,  3.52s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠇ Processed 28 chunks, 333 entities(duplicated), 290 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  80%|████████  | 28/35 [01:52<00:28,  4.08s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠏ Processed 29 chunks, 348 entities(duplicated), 307 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  83%|████████▎ | 29/35 [01:59<00:29,  4.88s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠋ Processed 30 chunks, 362 entities(duplicated), 329 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  86%|████████▌ | 30/35 [02:02<00:21,  4.29s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠙ Processed 31 chunks, 373 entities(duplicated), 337 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  89%|████████▊ | 31/35 [02:03<00:13,  3.28s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠹ Processed 32 chunks, 390 entities(duplicated), 369 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  91%|█████████▏| 32/35 [02:03<00:07,  2.55s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠸ Processed 33 chunks, 405 entities(duplicated), 378 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  94%|█████████▍| 33/35 [02:07<00:05,  2.84s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠼ Processed 34 chunks, 435 entities(duplicated), 395 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  97%|█████████▋| 34/35 [02:10<00:02,  2.94s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠴ Processed 35 chunks, 456 entities(duplicated), 440 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks: 100%|██████████| 35/35 [02:23<00:00,  4.10s/chunk]\u001b[A\n",
+      "INFO:lightrag:Inserting entities into storage...\n",
+      "\n",
+      "Inserting entities: 100%|██████████| 324/324 [00:00<00:00, 17456.96entity/s]\n",
+      "INFO:lightrag:Inserting relationships into storage...\n",
+      "\n",
+      "Inserting relationships: 100%|██████████| 427/427 [00:00<00:00, 29956.31relationship/s]\n",
+      "INFO:lightrag:Inserting 324 vectors to entities\n",
+      "\n",
+      "Generating embeddings:   0%|          | 0/11 [00:00<?, ?batch/s]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:   9%|▉         | 1/11 [00:00<00:06,  1.48batch/s]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:  18%|█▊        | 2/11 [00:02<00:11,  1.25s/batch]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:  27%|██▋       | 3/11 [00:02<00:06,  1.17batch/s]\u001b[A\n",
+      "Generating embeddings:  36%|███▋      | 4/11 [00:03<00:04,  1.50batch/s]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:  45%|████▌     | 5/11 [00:03<00:03,  1.78batch/s]\u001b[A\n",
+      "Generating embeddings:  55%|█████▍    | 6/11 [00:03<00:02,  2.01batch/s]\u001b[A\n",
+      "Generating embeddings:  64%|██████▎   | 7/11 [00:04<00:01,  2.19batch/s]\u001b[A\n",
+      "Generating embeddings:  73%|███████▎  | 8/11 [00:04<00:01,  2.31batch/s]\u001b[A\n",
+      "Generating embeddings:  82%|████████▏ | 9/11 [00:04<00:00,  2.41batch/s]\u001b[A\n",
+      "Generating embeddings:  91%|█████████ | 10/11 [00:05<00:00,  2.48batch/s]\u001b[A\n",
+      "Generating embeddings: 100%|██████████| 11/11 [00:05<00:00,  1.91batch/s]\u001b[A\n",
+      "INFO:lightrag:Inserting 427 vectors to relationships\n",
+      "\n",
+      "Generating embeddings:   0%|          | 0/14 [00:00<?, ?batch/s]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:   7%|▋         | 1/14 [00:01<00:14,  1.11s/batch]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:  14%|█▍        | 2/14 [00:02<00:14,  1.18s/batch]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:  21%|██▏       | 3/14 [00:02<00:08,  1.23batch/s]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:  29%|██▊       | 4/14 [00:03<00:06,  1.56batch/s]\u001b[A\n",
+      "Generating embeddings:  36%|███▌      | 5/14 [00:03<00:04,  1.85batch/s]\u001b[A\n",
+      "Generating embeddings:  43%|████▎     | 6/14 [00:03<00:03,  2.05batch/s]\u001b[A\n",
+      "Generating embeddings:  50%|█████     | 7/14 [00:04<00:03,  2.23batch/s]\u001b[A\n",
+      "Generating embeddings:  57%|█████▋    | 8/14 [00:04<00:02,  2.37batch/s]\u001b[A\n",
+      "Generating embeddings:  64%|██████▍   | 9/14 [00:04<00:02,  2.46batch/s]\u001b[A\n",
+      "Generating embeddings:  71%|███████▏  | 10/14 [00:05<00:01,  2.54batch/s]\u001b[A\n",
+      "Generating embeddings:  79%|███████▊  | 11/14 [00:05<00:01,  2.59batch/s]\u001b[A\n",
+      "Generating embeddings:  86%|████████▌ | 12/14 [00:06<00:00,  2.64batch/s]\u001b[A\n",
+      "Generating embeddings:  93%|█████████▎| 13/14 [00:06<00:00,  2.65batch/s]\u001b[A\n",
+      "Generating embeddings: 100%|██████████| 14/14 [00:06<00:00,  2.05batch/s]\u001b[A\n",
+      "INFO:lightrag:Writing graph with 333 nodes, 427 edges\n",
+      "Processing batch 1: 100%|██████████| 1/1 [02:45<00:00, 165.90s/it]\n"
+     ]
+    }
+   ],
+   "source": [
+    "# rag.insert(content)\n",
+    "rag.insert(content, split_by_character=\"\\n#\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "id": "c4f9ae517151a01d",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-01-09T03:45:11.668987Z",
+     "start_time": "2025-01-09T03:45:11.664744Z"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "prompt1 = \"\"\"你是一名经验丰富的论文分析科学家，你的任务是对一篇英文学术研究论文进行关键信息提取并深入分析。\n",
+    "请按照以下步骤进行分析：\n",
+    "1. 该文献主要研究的问题是什么？\n",
+    "2. 该文献采用什么方法进行分析？\n",
+    "3. 该文献的主要结论是什么？\n",
+    "首先在<分析>标签中，针对每个问题详��分析你的思考过程。然后在<回答>标签中给出所有问题的最终答案。\"\"\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "id": "7a6491385b050095",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-01-09T03:45:40.829111Z",
+     "start_time": "2025-01-09T03:45:13.530298Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:lightrag:Local query uses 5 entites, 12 relations, 3 text units\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:lightrag:Global query uses 8 entites, 5 relations, 4 text units\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "<分析>\n",
+      "1. **该文献主要研究的问题是什么？**\n",
+      "    - 思考过程：通过浏览论文内容，查找作者明确阐述研究目的的部分。文中多处提及“Our study was performed to explore whether folic acid treatment was associated with cancer outcomes and all-cause mortality after extended follow-up”，表明作者旨在探究叶酸治疗与癌症结局及全因死亡率之间的关系，尤其是在经过长期随访后。\n",
+      "2. **该文献采用什么方法进行分析？**\n",
+      "    - 思考过程：寻找描述研究方法和数据分析过程的段落。文中提到“Survival curves were constructed using the Kaplan-Meier method and differences in survival between groups were analyzed using the log-rank test. Estimates of hazard ratios (HRs) with 95% CIs were obtained by using Cox proportional hazards regression models stratified by trial”，可以看出作者使用了Kaplan-Meier法构建生存曲线、log-rank检验分析组间生存差异以及Cox比例风险回归模型估计风险比等方法。\n",
+      "3. **该文献的主要结论是什么？**\n",
+      "    - 思考过程：定位到论文中总结结论的部分，如“Conclusion Treatment with folic acid plus vitamin $\\mathsf{B}_{12}$ was associated with increased cancer outcomes and all-cause mortality in patients with ischemic heart disease in Norway, where there is no folic acid fortification of foods”，可知作者得出叶酸加维生素$\\mathsf{B}_{12}$治疗与癌症结局和全因死亡率增加有关的结论。\n",
+      "<回答>\n",
+      "1. 该文献主要研究的问题是：叶酸治疗与癌症结局及全因死亡率之间的关系，尤其是在经过长期随访后，叶酸治疗是否与癌症结局和全因死亡率相关。\n",
+      "2. 该文献采用的分析方法包括：使用Kaplan-Meier法构建生存曲线、log-rank检验分析组间生存差异、Cox比例风险回归模型估计风险比等。\n",
+      "3. 该文献的主要结论是：在挪威没有叶酸强化食品的情况下，叶酸加维生素$\\mathsf{B}_{12}$治疗与缺血性心脏病患者的癌症结局和全因死亡率增加有关。\n",
+      "\n",
+      "**参考文献**\n",
+      "- [VD] In2Norwegianhomocysteine-lowering trialsamongpatientswithischemicheart disease, there was a statistically nonsignificantincreaseincancerincidenceinthe groupsassignedtofolicacidtreatment.15,16 Our study was performed to explore whetherfolicacidtreatmentwasassociatedwithcanceroutcomesandall-cause mortality after extended follow-up.\n",
+      "- [VD] Survivalcurveswereconstructedusing theKaplan-Meiermethodanddifferences insurvivalbetweengroupswereanalyzed usingthelog-ranktest.Estimatesofhazard ratios (HRs) with $95\\%$ CIs were obtainedbyusingCoxproportionalhazards regressionmodelsstratifiedbytrial.\n",
+      "- [VD] Conclusion Treatment with folic acid plus vitamin $\\mathsf{B}_{12}$ was associated with increased cancer outcomes and all-cause mortality in patients with ischemic heart disease in Norway, where there is no folic acid fortification of foods.\n"
+     ]
+    }
+   ],
+   "source": [
+    "resp = rag.query(prompt1, param=QueryParam(mode=\"mix\", top_k=5))\n",
+    "print(resp)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "4e5bfad24cb721a8",
+   "metadata": {},
+   "source": "#### split by character only"
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "id": "44e2992dc95f8ce0",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-01-09T03:47:40.988796Z",
+     "start_time": "2025-01-09T03:47:40.982648Z"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "WORKING_DIR = \"../../llm_rag/paper_db/R000088_test2\"\n",
+    "if not os.path.exists(WORKING_DIR):\n",
+    "    os.mkdir(WORKING_DIR)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "id": "62c63385d2d973d5",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-01-09T03:51:39.951329Z",
+     "start_time": "2025-01-09T03:49:15.218976Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "INFO:lightrag:Logger initialized for working directory: ../../llm_rag/paper_db/R000088_test2\n",
+      "INFO:lightrag:Load KV llm_response_cache with 0 data\n",
+      "INFO:lightrag:Load KV full_docs with 0 data\n",
+      "INFO:lightrag:Load KV text_chunks with 0 data\n",
+      "INFO:nano-vectordb:Init {'embedding_dim': 4096, 'metric': 'cosine', 'storage_file': '../../llm_rag/paper_db/R000088_test2/vdb_entities.json'} 0 data\n",
+      "INFO:nano-vectordb:Init {'embedding_dim': 4096, 'metric': 'cosine', 'storage_file': '../../llm_rag/paper_db/R000088_test2/vdb_relationships.json'} 0 data\n",
+      "INFO:nano-vectordb:Init {'embedding_dim': 4096, 'metric': 'cosine', 'storage_file': '../../llm_rag/paper_db/R000088_test2/vdb_chunks.json'} 0 data\n",
+      "INFO:lightrag:Loaded document status storage with 0 records\n",
+      "INFO:lightrag:Processing 1 new unique documents\n",
+      "Processing batch 1:   0%|          | 0/1 [00:00<?, ?it/s]INFO:lightrag:Inserting 12 vectors to chunks\n",
+      "\n",
+      "Generating embeddings:   0%|          | 0/1 [00:00<?, ?batch/s]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings: 100%|██████████| 1/1 [00:02<00:00,  2.95s/batch]\u001b[A\n",
+      "\n",
+      "Extracting entities from chunks:   0%|          | 0/12 [00:00<?, ?chunk/s]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠙ Processed 1 chunks, 0 entities(duplicated), 0 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:   8%|▊         | 1/12 [00:03<00:43,  3.93s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠹ Processed 2 chunks, 8 entities(duplicated), 8 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  17%|█▋        | 2/12 [00:29<02:44, 16.46s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠸ Processed 3 chunks, 17 entities(duplicated), 15 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  25%|██▌       | 3/12 [00:30<01:25,  9.45s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠼ Processed 4 chunks, 27 entities(duplicated), 22 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  33%|███▎      | 4/12 [00:39<01:16,  9.52s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠴ Processed 5 chunks, 36 entities(duplicated), 33 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  42%|████▏     | 5/12 [00:40<00:43,  6.24s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠦ Processed 6 chunks, 49 entities(duplicated), 42 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  50%|█████     | 6/12 [00:49<00:43,  7.33s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠧ Processed 7 chunks, 62 entities(duplicated), 65 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  58%|█████▊    | 7/12 [01:05<00:50, 10.05s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠇ Processed 8 chunks, 81 entities(duplicated), 90 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  67%|██████▋   | 8/12 [01:23<00:50, 12.69s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠏ Processed 9 chunks, 99 entities(duplicated), 117 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  75%|███████▌  | 9/12 [01:32<00:34, 11.54s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠋ Processed 10 chunks, 123 entities(duplicated), 140 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  83%|████████▎ | 10/12 [01:48<00:25, 12.79s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠙ Processed 11 chunks, 158 entities(duplicated), 174 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks:  92%|█████████▏| 11/12 [02:03<00:13, 13.50s/chunk]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⠹ Processed 12 chunks, 194 entities(duplicated), 221 relations(duplicated)\r"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Extracting entities from chunks: 100%|██████████| 12/12 [02:13<00:00, 11.15s/chunk]\u001b[A\n",
+      "INFO:lightrag:Inserting entities into storage...\n",
+      "\n",
+      "Inserting entities: 100%|██████████| 170/170 [00:00<00:00, 11610.25entity/s]\n",
+      "INFO:lightrag:Inserting relationships into storage...\n",
+      "\n",
+      "Inserting relationships: 100%|██████████| 218/218 [00:00<00:00, 15913.51relationship/s]\n",
+      "INFO:lightrag:Inserting 170 vectors to entities\n",
+      "\n",
+      "Generating embeddings:   0%|          | 0/6 [00:00<?, ?batch/s]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:  17%|█▋        | 1/6 [00:01<00:05,  1.10s/batch]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:  33%|███▎      | 2/6 [00:02<00:04,  1.07s/batch]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:  50%|█████     | 3/6 [00:02<00:02,  1.33batch/s]\u001b[A\n",
+      "Generating embeddings:  67%|██████▋   | 4/6 [00:02<00:01,  1.67batch/s]\u001b[A\n",
+      "Generating embeddings:  83%|████████▎ | 5/6 [00:03<00:00,  1.95batch/s]\u001b[A\n",
+      "Generating embeddings: 100%|██████████| 6/6 [00:03<00:00,  1.66batch/s]\u001b[A\n",
+      "INFO:lightrag:Inserting 218 vectors to relationships\n",
+      "\n",
+      "Generating embeddings:   0%|          | 0/7 [00:00<?, ?batch/s]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:  14%|█▍        | 1/7 [00:01<00:10,  1.74s/batch]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:  29%|██▊       | 2/7 [00:02<00:05,  1.04s/batch]\u001b[A\n",
+      "Generating embeddings:  43%|████▎     | 3/7 [00:02<00:02,  1.35batch/s]\u001b[A\n",
+      "Generating embeddings:  57%|█████▋    | 4/7 [00:03<00:01,  1.69batch/s]\u001b[AINFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "\n",
+      "Generating embeddings:  71%|███████▏  | 5/7 [00:03<00:01,  1.96batch/s]\u001b[A\n",
+      "Generating embeddings:  86%|████████▌ | 6/7 [00:03<00:00,  2.17batch/s]\u001b[A\n",
+      "Generating embeddings: 100%|██████████| 7/7 [00:04<00:00,  1.68batch/s]\u001b[A\n",
+      "INFO:lightrag:Writing graph with 174 nodes, 218 edges\n",
+      "Processing batch 1: 100%|██████████| 1/1 [02:24<00:00, 144.69s/it]\n"
+     ]
+    }
+   ],
+   "source": [
+    "rag = LightRAG(\n",
+    "    working_dir=WORKING_DIR,\n",
+    "    llm_model_func=llm_model_func,\n",
+    "    embedding_func=EmbeddingFunc(\n",
+    "        embedding_dim=4096, max_token_size=8192, func=embedding_func\n",
+    "    ),\n",
+    "    chunk_token_size=512,\n",
+    ")\n",
+    "\n",
+    "# rag.insert(content)\n",
+    "rag.insert(content, split_by_character=\"\\n#\", split_by_character_only=True)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "id": "3c7aa9836d8d43c7",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-01-09T03:52:37.000418Z",
+     "start_time": "2025-01-09T03:52:09.933584Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:lightrag:Local query uses 5 entites, 3 relations, 2 text units\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/embeddings \"HTTP/1.1 200 OK\"\n",
+      "INFO:lightrag:Global query uses 9 entites, 5 relations, 4 text units\n",
+      "INFO:httpx:HTTP Request: POST https://ark.cn-beijing.volces.com/api/v3/chat/completions \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "<分析>\n",
+      "- **该文献主要研究的问题是什么？**\n",
+      "    - **思考过程**：通过浏览论文的标题、摘要、引言等部分，寻找关于研究目的和问题的描述。论文标题为“Cancer Incidence and Mortality After Treatment With Folic Acid and Vitamin B12”，摘要中的“Objective”部分明确指出研究目的是“To evaluate effects of treatment with B vitamins on cancer outcomes and all-cause mortality in 2 randomized controlled trials”。因此，可以确定该文献主要研究的问题是评估B族维生素治疗对两项随机对照试验中癌症结局和全因死亡率的影响。\n",
+      "- **该文献采用什么方法进行分析？**\n",
+      "    - **思考过程**：在论文的“METHODS”部分详细描述了研究方法。文中提到这是一个对两项随机、双盲、安慰剂对照临床试验（Norwegian Vitamin [NORVIT] trial和Western Norway B Vitamin Intervention Trial [WENBIT]）数据的联合分析，并进行了观察性的试验后随访。具体包括对参与者进行分组干预（不同剂量的叶酸、维生素B12、维生素B6或安慰剂），收集临床信息和血样，分析循环B族维生素、同型半胱氨酸和可替宁等指标，并进行基因分型等，还涉及到多种统计分析方法，如计算预期癌症发生率、构建生存曲线、进行Cox比例风险回归模型分析等。\n",
+      "- **该文献的主要结论是什么？**\n",
+      "    - **思考过程**：在论文的“Results”和“Conclusion”部分寻找主要结论。研究结果表明，在治疗期间，接受叶酸加维生素B12治疗的参与者血清叶酸浓度显著增加，且在后续随访中，该组癌症发病率、癌症死亡率和全因死亡率均有所上升，主要是肺癌发病率增加，而维生素B6治疗未显示出显著影响。结论部分明确指出“Treatment with folic acid plus vitamin $\\mathsf{B}_{12}$ was associated with increased cancer outcomes and all-cause mortality in patients with ischemic heart disease in Norway, where there is no folic acid fortification of foods”。\n",
+      "</分析>\n",
+      "\n",
+      "<回答>\n",
+      "- **主要研究问题**：评估B族维生素治疗对两项随机对照试验中癌症结局和全因死亡率的影响。\n",
+      "- **研究方法**：采用对两项随机、双盲、安慰剂对照临床试验（Norwegian Vitamin [NORVIT] trial和Western Norway B Vitamin Intervention Trial [WENBIT]）数据的联合分析，并进行观察性的试验后随访，涉及分组干预、多种指标检测以及多种统计分析方法。\n",
+      "- **主要结论**：在挪威（食品中未添加叶酸），对于缺血性心脏病患者，叶酸加维生素B12治疗与癌症结局和全因死亡率的增加有关，而维生素B6治疗未显示出显著影响。\n",
+      "\n",
+      "**参考文献**\n",
+      "- [VD] Cancer Incidence and Mortality After Treatment With Folic Acid and Vitamin B12\n",
+      "- [VD] METHODS Study Design, Participants, and Study Intervention\n",
+      "- [VD] RESULTS\n",
+      "- [VD] Conclusion\n",
+      "- [VD] Objective To evaluate effects of treatment with B vitamins on cancer outcomes and all-cause mortality in 2 randomized controlled trials.\n"
+     ]
+    }
+   ],
+   "source": [
+    "resp = rag.query(prompt1, param=QueryParam(mode=\"mix\", top_k=5))\n",
+    "print(resp)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "7ba6fa79a2550d10",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 2
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython2",
+   "version": "2.7.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

lightrag/__init__.py

@@ -1,5 +1,5 @@
 from .lightrag import LightRAG as LightRAG, QueryParam as QueryParam
 
-__version__ = "1.0.9"
+__version__ = "1.1.0"
 __author__ = "Zirui Guo"
 __url__ = "https://github.com/HKUDS/LightRAG"

lightrag/lightrag.py

@@ -45,6 +45,7 @@ from .storage import (
 
 from .prompt import GRAPH_FIELD_SEP
 
+
 # future KG integrations
 
 # from .kg.ArangoDB_impl import (
@@ -167,7 +168,7 @@ class LightRAG:
 
     # LLM
     llm_model_func: callable = gpt_4o_mini_complete  # hf_model_complete#
-    llm_model_name: str = "meta-llama/Llama-3.2-1B-Instruct"  #'meta-llama/Llama-3.2-1B'#'google/gemma-2-2b-it'
+    llm_model_name: str = "meta-llama/Llama-3.2-1B-Instruct"  # 'meta-llama/Llama-3.2-1B'#'google/gemma-2-2b-it'
     llm_model_max_token_size: int = 32768
     llm_model_max_async: int = 16
     llm_model_kwargs: dict = field(default_factory=dict)
@@ -177,7 +178,7 @@ class LightRAG:
 
     enable_llm_cache: bool = True
     # Sometimes there are some reason the LLM failed at Extracting Entities, and we want to continue without LLM cost, we can use this flag
-    enable_llm_cache_for_entity_extract: bool = False
+    enable_llm_cache_for_entity_extract: bool = True
 
     # extension
     addon_params: dict = field(default_factory=dict)
@@ -186,6 +187,10 @@ class LightRAG:
     # Add new field for document status storage type
     doc_status_storage: str = field(default="JsonDocStatusStorage")
 
+    # Custom Chunking Function
+    chunking_func: callable = chunking_by_token_size
+    chunking_func_kwargs: dict = field(default_factory=dict)
+
     def __post_init__(self):
         log_file = os.path.join("lightrag.log")
         set_logger(log_file)
@@ -313,15 +318,25 @@ class LightRAG:
             "JsonDocStatusStorage": JsonDocStatusStorage,
         }
 
-    def insert(self, string_or_strings):
+    def insert(
+        self, string_or_strings, split_by_character=None, split_by_character_only=False
+    ):
         loop = always_get_an_event_loop()
-        return loop.run_until_complete(self.ainsert(string_or_strings))
+        return loop.run_until_complete(
+            self.ainsert(string_or_strings, split_by_character, split_by_character_only)
+        )
 
-    async def ainsert(self, string_or_strings):
+    async def ainsert(
+        self, string_or_strings, split_by_character=None, split_by_character_only=False
+    ):
         """Insert documents with checkpoint support
 
         Args:
             string_or_strings: Single document string or list of document strings
+            split_by_character: if split_by_character is not None, split the string by character, if chunk longer than
+            chunk_size, split the sub chunk by token size.
+            split_by_character_only: if split_by_character_only is True, split the string by character only, when
+            split_by_character is None, this parameter is ignored.
         """
         if isinstance(string_or_strings, str):
             string_or_strings = [string_or_strings]
@@ -358,7 +373,7 @@ class LightRAG:
             batch_docs = dict(list(new_docs.items())[i : i + batch_size])
 
             for doc_id, doc in tqdm_async(
-                batch_docs.items(), desc=f"Processing batch {i//batch_size + 1}"
+                batch_docs.items(), desc=f"Processing batch {i // batch_size + 1}"
             ):
                 try:
                     # Update status to processing
@@ -377,11 +392,14 @@ class LightRAG:
                             **dp,
                             "full_doc_id": doc_id,
                         }
-                        for dp in chunking_by_token_size(
+                        for dp in self.chunking_func(
                             doc["content"],
+                            split_by_character=split_by_character,
+                            split_by_character_only=split_by_character_only,
                             overlap_token_size=self.chunk_overlap_token_size,
                             max_token_size=self.chunk_token_size,
                             tiktoken_model=self.tiktoken_model_name,
+                            **self.chunking_func_kwargs,
                         )
                     }
 
@@ -453,6 +471,73 @@ class LightRAG:
                     # Ensure all indexes are updated after each document
                     await self._insert_done()
 
+    def insert_custom_chunks(self, full_text: str, text_chunks: list[str]):
+        loop = always_get_an_event_loop()
+        return loop.run_until_complete(
+            self.ainsert_custom_chunks(full_text, text_chunks)
+        )
+
+    async def ainsert_custom_chunks(self, full_text: str, text_chunks: list[str]):
+        update_storage = False
+        try:
+            doc_key = compute_mdhash_id(full_text.strip(), prefix="doc-")
+            new_docs = {doc_key: {"content": full_text.strip()}}
+
+            _add_doc_keys = await self.full_docs.filter_keys([doc_key])
+            new_docs = {k: v for k, v in new_docs.items() if k in _add_doc_keys}
+            if not len(new_docs):
+                logger.warning("This document is already in the storage.")
+                return
+
+            update_storage = True
+            logger.info(f"[New Docs] inserting {len(new_docs)} docs")
+
+            inserting_chunks = {}
+            for chunk_text in text_chunks:
+                chunk_text_stripped = chunk_text.strip()
+                chunk_key = compute_mdhash_id(chunk_text_stripped, prefix="chunk-")
+
+                inserting_chunks[chunk_key] = {
+                    "content": chunk_text_stripped,
+                    "full_doc_id": doc_key,
+                }
+
+            _add_chunk_keys = await self.text_chunks.filter_keys(
+                list(inserting_chunks.keys())
+            )
+            inserting_chunks = {
+                k: v for k, v in inserting_chunks.items() if k in _add_chunk_keys
+            }
+            if not len(inserting_chunks):
+                logger.warning("All chunks are already in the storage.")
+                return
+
+            logger.info(f"[New Chunks] inserting {len(inserting_chunks)} chunks")
+
+            await self.chunks_vdb.upsert(inserting_chunks)
+
+            logger.info("[Entity Extraction]...")
+            maybe_new_kg = await extract_entities(
+                inserting_chunks,
+                knowledge_graph_inst=self.chunk_entity_relation_graph,
+                entity_vdb=self.entities_vdb,
+                relationships_vdb=self.relationships_vdb,
+                global_config=asdict(self),
+            )
+
+            if maybe_new_kg is None:
+                logger.warning("No new entities and relationships found")
+                return
+            else:
+                self.chunk_entity_relation_graph = maybe_new_kg
+
+            await self.full_docs.upsert(new_docs)
+            await self.text_chunks.upsert(inserting_chunks)
+
+        finally:
+            if update_storage:
+                await self._insert_done()
+
     async def _insert_done(self):
         tasks = []
         for storage_inst in [

lightrag/operate.py

@@ -4,7 +4,6 @@ import re
 from tqdm.asyncio import tqdm as tqdm_async
 from typing import Union
 from collections import Counter, defaultdict
-import warnings
 from .utils import (
     logger,
     clean_str,
@@ -34,23 +33,61 @@ import time
 
 
 def chunking_by_token_size(
-    content: str, overlap_token_size=128, max_token_size=1024, tiktoken_model="gpt-4o"
+    content: str,
+    split_by_character=None,
+    split_by_character_only=False,
+    overlap_token_size=128,
+    max_token_size=1024,
+    tiktoken_model="gpt-4o",
+    **kwargs,
 ):
     tokens = encode_string_by_tiktoken(content, model_name=tiktoken_model)
     results = []
-    for index, start in enumerate(
-        range(0, len(tokens), max_token_size - overlap_token_size)
-    ):
-        chunk_content = decode_tokens_by_tiktoken(
-            tokens[start : start + max_token_size], model_name=tiktoken_model
-        )
-        results.append(
-            {
-                "tokens": min(max_token_size, len(tokens) - start),
-                "content": chunk_content.strip(),
-                "chunk_order_index": index,
-            }
-        )
+    if split_by_character:
+        raw_chunks = content.split(split_by_character)
+        new_chunks = []
+        if split_by_character_only:
+            for chunk in raw_chunks:
+                _tokens = encode_string_by_tiktoken(chunk, model_name=tiktoken_model)
+                new_chunks.append((len(_tokens), chunk))
+        else:
+            for chunk in raw_chunks:
+                _tokens = encode_string_by_tiktoken(chunk, model_name=tiktoken_model)
+                if len(_tokens) > max_token_size:
+                    for start in range(
+                        0, len(_tokens), max_token_size - overlap_token_size
+                    ):
+                        chunk_content = decode_tokens_by_tiktoken(
+                            _tokens[start : start + max_token_size],
+                            model_name=tiktoken_model,
+                        )
+                        new_chunks.append(
+                            (min(max_token_size, len(_tokens) - start), chunk_content)
+                        )
+                else:
+                    new_chunks.append((len(_tokens), chunk))
+        for index, (_len, chunk) in enumerate(new_chunks):
+            results.append(
+                {
+                    "tokens": _len,
+                    "content": chunk.strip(),
+                    "chunk_order_index": index,
+                }
+            )
+    else:
+        for index, start in enumerate(
+            range(0, len(tokens), max_token_size - overlap_token_size)
+        ):
+            chunk_content = decode_tokens_by_tiktoken(
+                tokens[start : start + max_token_size], model_name=tiktoken_model
+            )
+            results.append(
+                {
+                    "tokens": min(max_token_size, len(tokens) - start),
+                    "content": chunk_content.strip(),
+                    "chunk_order_index": index,
+                }
+            )
     return results
 
 
@@ -574,15 +611,22 @@ async def kg_query(
         logger.warning("low_level_keywords and high_level_keywords is empty")
         return PROMPTS["fail_response"]
     if ll_keywords == [] and query_param.mode in ["local", "hybrid"]:
-        logger.warning("low_level_keywords is empty")
-        return PROMPTS["fail_response"]
-    else:
-        ll_keywords = ", ".join(ll_keywords)
+        logger.warning(
+            "low_level_keywords is empty, switching from %s mode to global mode",
+            query_param.mode,
+        )
+        query_param.mode = "global"
     if hl_keywords == [] and query_param.mode in ["global", "hybrid"]:
-        logger.warning("high_level_keywords is empty")
-        return PROMPTS["fail_response"]
-    else:
-        hl_keywords = ", ".join(hl_keywords)
+        logger.warning(
+            "high_level_keywords is empty, switching from %s mode to local mode",
+            query_param.mode,
+        )
+        query_param.mode = "local"
+
+    ll_keywords = ", ".join(ll_keywords) if ll_keywords else ""
+    hl_keywords = ", ".join(hl_keywords) if hl_keywords else ""
+
+    logger.info("Using %s mode for query processing", query_param.mode)
 
     # Build context
     keywords = [ll_keywords, hl_keywords]
@@ -648,77 +692,51 @@ async def _build_query_context(
     # ll_entities_context, ll_relations_context, ll_text_units_context = "", "", ""
     # hl_entities_context, hl_relations_context, hl_text_units_context = "", "", ""
 
-    ll_kewwords, hl_keywrds = query[0], query[1]
-    if query_param.mode in ["local", "hybrid"]:
-        if ll_kewwords == "":
-            ll_entities_context, ll_relations_context, ll_text_units_context = (
-                "",
-                "",
-                "",
-            )
-            warnings.warn(
-                "Low Level context is None. Return empty Low entity/relationship/source"
-            )
-            query_param.mode = "global"
-        else:
-            (
-                ll_entities_context,
-                ll_relations_context,
-                ll_text_units_context,
-            ) = await _get_node_data(
-                ll_kewwords,
-                knowledge_graph_inst,
-                entities_vdb,
-                text_chunks_db,
-                query_param,
-            )
-    if query_param.mode in ["global", "hybrid"]:
-        if hl_keywrds == "":
-            hl_entities_context, hl_relations_context, hl_text_units_context = (
-                "",
-                "",
-                "",
-            )
-            warnings.warn(
-                "High Level context is None. Return empty High entity/relationship/source"
-            )
-            query_param.mode = "local"
-        else:
-            (
-                hl_entities_context,
-                hl_relations_context,
-                hl_text_units_context,
-            ) = await _get_edge_data(
-                hl_keywrds,
-                knowledge_graph_inst,
-                relationships_vdb,
-                text_chunks_db,
-                query_param,
-            )
-            if (
-                hl_entities_context == ""
-                and hl_relations_context == ""
-                and hl_text_units_context == ""
-            ):
-                logger.warn("No high level context found. Switching to local mode.")
-                query_param.mode = "local"
-    if query_param.mode == "hybrid":
-        entities_context, relations_context, text_units_context = combine_contexts(
-            [hl_entities_context, ll_entities_context],
-            [hl_relations_context, ll_relations_context],
-            [hl_text_units_context, ll_text_units_context],
+    ll_keywords, hl_keywords = query[0], query[1]
+
+    if query_param.mode == "local":
+        entities_context, relations_context, text_units_context = await _get_node_data(
+            ll_keywords,
+            knowledge_graph_inst,
+            entities_vdb,
+            text_chunks_db,
+            query_param,
+        )
+    elif query_param.mode == "global":
+        entities_context, relations_context, text_units_context = await _get_edge_data(
+            hl_keywords,
+            knowledge_graph_inst,
+            relationships_vdb,
+            text_chunks_db,
+            query_param,
         )
-    elif query_param.mode == "local":
-        entities_context, relations_context, text_units_context = (
+    else:  # hybrid mode
+        (
             ll_entities_context,
             ll_relations_context,
             ll_text_units_context,
+        ) = await _get_node_data(
+            ll_keywords,
+            knowledge_graph_inst,
+            entities_vdb,
+            text_chunks_db,
+            query_param,
         )
-    elif query_param.mode == "global":
-        entities_context, relations_context, text_units_context = (
+        (
             hl_entities_context,
             hl_relations_context,
             hl_text_units_context,
+        ) = await _get_edge_data(
+            hl_keywords,
+            knowledge_graph_inst,
+            relationships_vdb,
+            text_chunks_db,
+            query_param,
+        )
+        entities_context, relations_context, text_units_context = combine_contexts(
+            [hl_entities_context, ll_entities_context],
+            [hl_relations_context, ll_relations_context],
+            [hl_text_units_context, ll_text_units_context],
         )
     return f"""
 -----Entities-----

requirements.txt

@@ -1,38 +1,38 @@
 accelerate
-aioboto3~=13.3.0
-aiofiles~=24.1.0
-aiohttp~=3.11.11
-asyncpg~=0.30.0
+aioboto3
+aiofiles
+aiohttp
+asyncpg
 
 # database packages
 graspologic
 gremlinpython
 hnswlib
 nano-vectordb
-neo4j~=5.27.0
-networkx~=3.2.1
+neo4j
+networkx
 
-numpy~=2.2.0
-ollama~=0.4.4
-openai~=1.58.1
+numpy
+ollama
+openai
 oracledb
-psycopg-pool~=3.2.4
-psycopg[binary,pool]~=3.2.3
-pydantic~=2.10.4
+psycopg-pool
+psycopg[binary,pool]
+pydantic
 pymilvus
 pymongo
 pymysql
-python-dotenv~=1.0.1
-pyvis~=0.3.2
-setuptools~=70.0.0
+python-dotenv
+pyvis
+setuptools
 # lmdeploy[all]
-sqlalchemy~=2.0.36
-tenacity~=9.0.0
+sqlalchemy
+tenacity
 
 
 # LLM packages
-tiktoken~=0.8.0
-torch~=2.5.1+cu121
-tqdm~=4.67.1
-transformers~=4.47.1
+tiktoken
+torch
+tqdm
+transformers
 xxhash