Spaces:
Sleeping
Sleeping
File size: 12,073 Bytes
5fdb69e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 |
{
"cells": [
{
"cell_type": "markdown",
"id": "f3ce7a00-62c7-4cee-bed6-a89bf052e167",
"metadata": {},
"source": [
"# Colab Notebook RAG Assistant\n",
"\n",
"Short Notebook Description:\n",
"\n",
"This Jupyter/Colab notebook builds a Retrieval-Augmented Generation (RAG) assistant over your own collection of .ipynb files in Google Colab. It:\n",
"\n",
"1. Loads all notebooks from a local folder or mounted Google Drive.\n",
"\n",
"2. Chunks their content into manageable pieces.\n",
"\n",
"3. Embeds each chunk with OpenAI embeddings and stores them in a persistent Chroma vector database.\n",
"\n",
"4. Provides a ConversationalRetrievalChain with memory and a Gradio chat interface.\n",
"\n",
"5. For any user question, it returns both an answer and the names of the exact notebooks where the relevant information was found.\n",
"\n",
"This setup lets you query your entire notebook history—whether local or in Colab—just like a personal knowledge base."
]
},
{
"cell_type": "markdown",
"id": "1a7a0225-800c-4088-9bd2-ac98dbbb55c9",
"metadata": {},
"source": [
"## Imports"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "fe7e9772-171f-4ff6-bd3b-e77aa82b19d3",
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"import glob\n",
"from dotenv import load_dotenv\n",
"import gradio as gr\n",
"\n",
"from langchain_openai import OpenAIEmbeddings, ChatOpenAI\n",
"from langchain.document_loaders import DirectoryLoader, NotebookLoader\n",
"from langchain.text_splitter import CharacterTextSplitter, RecursiveCharacterTextSplitter\n",
"from langchain_chroma import Chroma\n",
"from langchain.memory import ConversationBufferMemory\n",
"from langchain.chains import ConversationalRetrievalChain\n",
"from langchain.chains import RetrievalQA\n",
"\n",
"from sklearn.manifold import TSNE\n",
"import plotly.graph_objects as go\n",
"import matplotlib.pyplot as plt\n",
"\n",
"import numpy as np"
]
},
{
"cell_type": "markdown",
"id": "8dfd2b57-3b3b-4fc2-bb2d-40be7aba3a4a",
"metadata": {},
"source": [
"## Configuration & Set Environment"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "5d502b28-1d33-43bc-8797-41fed26d5fa0",
"metadata": {},
"outputs": [],
"source": [
"load_dotenv(override = True)\n",
"\n",
"OPENAI_KEY = os.getenv('OPENAI_API_KEY')\n",
"NOTEBOOKS_DIR = os.getenv('NOTEBOOKS_DIR')\n",
"VECTOR_DB_DIR = os.getenv('VECTOR_DB_DIR')"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "af7f5fa2-78f8-45bf-a0e0-e1f93cc98f4b",
"metadata": {},
"outputs": [],
"source": [
"MODEL = 'gpt-4o-mini'\n",
"CHUNK_SIZE = 1000\n",
"CHUNK_OVERLAP = 200"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "82f7b583-d176-448b-b762-28618f05c660",
"metadata": {},
"outputs": [],
"source": [
"try:\n",
" # Colab\n",
" from google.colab import drive\n",
" print(\"Running in Colab: mounting Google Drive...\")\n",
" drive.mount('/content/drive')\n",
" is_colab = True\n",
"\n",
" # Colab defaults\n",
" NOTEBOOKS_DIR = '/content/drive/MyDrive/ColabNotebooks'\n",
" DB_DIR = VECTOR_DB_DIR or '/content/drive/MyDrive/colab_vector_db'\n",
"\n",
"except ImportError:\n",
" # Local Jupyter Lab:\n",
" print(\"Not in Colab: using local notebooks directory.\")\n",
" NOTEBOOKS_DIR = os.path.expanduser(NOTEBOOKS_DIR)\n",
" DB_DIR = VECTOR_DB_DIR\n",
"\n",
" # Verify the local notebooks directory exists\n",
" if not os.path.isdir(NOTEBOOKS_DIR):\n",
" raise FileNotFoundError(\n",
" f\"Local notebooks directory '{NOTEBOOKS_DIR}' not found.\" \n",
" \"\\nPlease sync your Google Drive folder (e.g., via Drive for Desktop) \"\n",
" \"or set NOTEBOOKS_DIR in your .env to the correct path.\"\n",
" )\n",
"# Confirm final paths\n",
"# print(f\"Indexing notebooks from: {NOTEBOOKS_DIR}\")\n",
"# print(f\"Chroma will store embeddings in: {DB_DIR}\")"
]
},
{
"cell_type": "markdown",
"id": "5eefd329-712f-4c43-b7aa-0322c0cd7c41",
"metadata": {},
"source": [
"## Read in Notebook files"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "c4468cbc-8f04-47c7-a583-1cb81cdb17fb",
"metadata": {},
"outputs": [],
"source": [
"notebooks = glob.glob(\n",
" os.path.join(NOTEBOOKS_DIR, \"**\", \"*.ipynb\"),\n",
" recursive=True\n",
")\n",
"print(f\"Notebooks found: {len(notebooks)}\")\n",
"\n",
"\n",
"loader = DirectoryLoader(NOTEBOOKS_DIR,\n",
" glob = '**/*.ipynb',\n",
" loader_cls = NotebookLoader\n",
" )\n",
"docs = loader.load()"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "3f8c0ce3-e3a9-4271-9824-e7fa42c8867d",
"metadata": {},
"outputs": [],
"source": [
"splitter = RecursiveCharacterTextSplitter(\n",
" chunk_size = CHUNK_SIZE, \n",
" chunk_overlap = CHUNK_OVERLAP, \n",
" separators=[\"\\n## \", \"\\n### \", \"\\n#### \", \"\\n\\n\", \"\\n\", \" \", \"\"]\n",
")\n",
"\n",
"chunks = splitter.split_documents(docs)\n",
"print(f'Created {len(chunks)} chunks from your notebooks')"
]
},
{
"cell_type": "markdown",
"id": "d73f8869-020b-48ac-bce9-82fadd58e04b",
"metadata": {},
"source": [
"## Embedding"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "27d6269c-88ac-4da7-8e87-691308d9e473",
"metadata": {},
"outputs": [],
"source": [
"embeddings = OpenAIEmbeddings(openai_api_key=OPENAI_KEY)\n",
"\n",
"if os.path.exists(DB_DIR):\n",
" Chroma(persist_directory = DB_DIR, embedding_function = embeddings).delete_collection()\n",
"\n",
"\n",
"vectorstore = Chroma.from_documents(\n",
" documents = chunks,\n",
" embedding = embeddings,\n",
" persist_directory = VECTOR_DB_DIR\n",
")\n",
"\n",
"vector_count = vectorstore._collection.count()\n",
"print(f\"Vectorstore contains {vector_count} vectors.\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "8d795bc7-82dc-4ad5-be39-97e08c033a4c",
"metadata": {},
"outputs": [],
"source": [
"sample_embedding = vectorstore._collection.get(limit=1, include=[\"embeddings\"])[\"embeddings\"][0]\n",
"dimensions = len(sample_embedding)\n",
"print(f\"There are {vectorstore._collection.count():,} vectors with {dimensions:,} dimensions in the vector store.\")"
]
},
{
"cell_type": "markdown",
"id": "bae1ab40-c22d-4815-bec4-840d69cf702b",
"metadata": {},
"source": [
"## Visualize in 3D"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "e3e30379-eb8f-469d-841e-cf95a542595b",
"metadata": {},
"outputs": [],
"source": [
"result = vectorstore._collection.get(include=['embeddings', 'documents',])\n",
"vectors = np.array(result['embeddings'])\n",
"documents = result['documents']\n",
"colors = ['blue'] * len(vectors)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "1b0c61e2-1d5d-429d-ace7-e883e051fdd2",
"metadata": {},
"outputs": [],
"source": [
"tsne = TSNE(n_components=3, random_state=42)\n",
"reduced_vectors = tsne.fit_transform(vectors)\n",
"\n",
"fig = go.Figure(data=[go.Scatter3d(\n",
" x=reduced_vectors[:, 0],\n",
" y=reduced_vectors[:, 1],\n",
" z=reduced_vectors[:, 2],\n",
" mode='markers',\n",
" marker=dict(size=4, color=colors, opacity=0.8),\n",
" text=[d[:100] + \"...\" for d in documents],\n",
" hoverinfo='text'\n",
")])\n",
"\n",
"fig.update_layout(\n",
" title='3D TSNE of Notebook-Chunks',\n",
" scene=dict(\n",
" xaxis_title=\"TSNE-1\",\n",
" yaxis_title=\"TSNE-2\",\n",
" zaxis_title=\"TSNE-3\"\n",
" ),\n",
" width=800,\n",
" height=600,\n",
" margin=dict(r=10, b=10, l=10, t=40)\n",
")\n",
"\n",
"fig.show()"
]
},
{
"cell_type": "markdown",
"id": "831684d5-5694-488f-aaed-e219d57b909c",
"metadata": {},
"source": [
"## Build LLM"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "02197e43-c958-4f70-be38-666ee4c1c4ae",
"metadata": {},
"outputs": [],
"source": [
"llm = ChatOpenAI(model_name = MODEL, temperature = 0.6)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "3aabba9e-e447-4597-a86e-fe3fc5b8babe",
"metadata": {},
"outputs": [],
"source": [
"qa = RetrievalQA.from_chain_type(\n",
" llm = llm,\n",
" chain_type=\"stuff\",\n",
" retriever=vectorstore.as_retriever(search_kwargs={\"k\": 4}),\n",
" return_source_documents=True\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "a71daee6-7d82-4212-ae4f-a2553f1d3c8a",
"metadata": {},
"outputs": [],
"source": [
"memory = ConversationBufferMemory(\n",
" memory_key = 'chat_history',\n",
" return_messages = True\n",
")\n",
"\n",
"conv_chain = ConversationalRetrievalChain.from_llm(\n",
" llm = llm,\n",
" retriever = vectorstore.as_retriever(search_kwargs = {'k': 10}),\n",
" memory = memory\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "8d8e2b9e-bb50-4bda-9832-c6a2779526e3",
"metadata": {},
"outputs": [],
"source": [
"def chat_with_memory_and_sources(message, chat_history):\n",
" # Get the conversational answer (memory included)\n",
" conv_res = conv_chain.invoke({\n",
" \"question\": message,\n",
" \"chat_history\": chat_history\n",
" })\n",
" answer = conv_res[\"answer\"]\n",
"\n",
" # Retrieve source documents \n",
" src_res = qa({\"query\": message})\n",
" src_docs = src_res[\"source_documents\"]\n",
"\n",
" # Extract and dedupe notebook filenames from metadata\n",
" notebooks = [\n",
" os.path.basename(doc.metadata.get(\"source\", \"\"))\n",
" for doc in src_docs\n",
" if doc.metadata.get(\"source\")\n",
" ]\n",
" unique = []\n",
" for n in notebooks:\n",
" if n not in unique:\n",
" unique.append(n)\n",
"\n",
" # Append the list of notebook filenames\n",
" if unique:\n",
" answer += \"\\n\\n**Found Notebooks:**\\n\" + \"\\n\".join(f\"- {n}\" for n in unique)\n",
" else:\n",
" answer += \"\\n\\n_No Notebooks found._\"\n",
"\n",
" return answer"
]
},
{
"cell_type": "markdown",
"id": "3d5c53ec-4fe3-46cc-9c17-7326294d24ef",
"metadata": {},
"source": [
"## Gradio UI"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "371e32ee-df20-4ec5-91eb-5023fc4b70b2",
"metadata": {},
"outputs": [],
"source": [
"view = gr.ChatInterface(chat_with_memory_and_sources, \n",
" title=\"Notebook-RAG-Assistant mit Memory & Quellen\",\n",
" type = 'messages').launch(inbrowser=True)"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.11.11"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
|