Create a comprehensive sample configuration file.
()
| 252 | |
| 253 | |
| 254 | def create_sample_config(): |
| 255 | """Create a comprehensive sample configuration file.""" |
| 256 | sample_config = { |
| 257 | "description": "Demo batch indexing configuration showcasing various features", |
| 258 | "pipeline_settings": { |
| 259 | "embedding_model_name": "Qwen/Qwen3-Embedding-0.6B", |
| 260 | "indexing": { |
| 261 | "embedding_batch_size": 50, |
| 262 | "enrichment_batch_size": 25, |
| 263 | "enable_progress_tracking": True |
| 264 | }, |
| 265 | "contextual_enricher": { |
| 266 | "enabled": True, |
| 267 | "window_size": 2, |
| 268 | "model_name": "qwen3:0.6b" |
| 269 | }, |
| 270 | "chunking": { |
| 271 | "chunk_size": 512, |
| 272 | "chunk_overlap": 64, |
| 273 | "enable_latechunk": True, |
| 274 | "enable_docling": True |
| 275 | }, |
| 276 | "retrievers": { |
| 277 | "dense": { |
| 278 | "enabled": True, |
| 279 | "lancedb_table_name": "demo_text_pages" |
| 280 | }, |
| 281 | "bm25": { |
| 282 | "enabled": True, |
| 283 | "index_name": "demo_bm25_index" |
| 284 | } |
| 285 | }, |
| 286 | "storage": { |
| 287 | "lancedb_uri": "./index_store/lancedb", |
| 288 | "bm25_path": "./index_store/bm25" |
| 289 | } |
| 290 | }, |
| 291 | "ollama_config": { |
| 292 | "generation_model": "qwen3:0.6b", |
| 293 | "embedding_model": "qwen3:0.6b" |
| 294 | }, |
| 295 | "indexes": [ |
| 296 | { |
| 297 | "name": "Sample Invoice Collection", |
| 298 | "description": "Demo index containing sample invoice documents", |
| 299 | "documents": [ |
| 300 | "./rag_system/documents/invoice_1039.pdf", |
| 301 | "./rag_system/documents/invoice_1041.pdf" |
| 302 | ], |
| 303 | "processing_options": { |
| 304 | "chunk_size": 512, |
| 305 | "enable_enrichment": True, |
| 306 | "retrieval_mode": "hybrid" |
| 307 | } |
| 308 | }, |
| 309 | { |
| 310 | "name": "Research Papers Demo", |
| 311 | "description": "Demo index for research papers and whitepapers", |