()
| 202 | return index_name |
| 203 | |
| 204 | def main(): |
| 205 | parser = argparse.ArgumentParser(description="Build a FAISS index from an Arrow datastore") |
| 206 | |
| 207 | # Only required parameter is the dstore_path |
| 208 | parser.add_argument( |
| 209 | "--dstore_path", |
| 210 | type=str, |
| 211 | required=True, |
| 212 | help="Path to the Arrow file containing the datastore" |
| 213 | ) |
| 214 | |
| 215 | # Optional parameters |
| 216 | parser.add_argument( |
| 217 | "--num_keys_to_add_at_a_time", |
| 218 | type=int, |
| 219 | default=1_000_000, |
| 220 | help="Number of keys to add at a time" |
| 221 | ) |
| 222 | parser.add_argument( |
| 223 | "--ncentroids", |
| 224 | type=int, |
| 225 | default=4096, |
| 226 | help="Number of centroids for IVFPQ" |
| 227 | ) |
| 228 | parser.add_argument( |
| 229 | "--seed", |
| 230 | type=int, |
| 231 | default=42, |
| 232 | help="Random seed" |
| 233 | ) |
| 234 | parser.add_argument( |
| 235 | "--code_size", |
| 236 | type=int, |
| 237 | default=32, |
| 238 | help="Code size for PQ" |
| 239 | ) |
| 240 | parser.add_argument( |
| 241 | "--probe", |
| 242 | type=int, |
| 243 | default=8, |
| 244 | help="Number of probes for query" |
| 245 | ) |
| 246 | |
| 247 | args = parser.parse_args() |
| 248 | |
| 249 | index_path = build_index( |
| 250 | dstore_path=args.dstore_path, |
| 251 | num_keys_to_add_at_a_time=args.num_keys_to_add_at_a_time, |
| 252 | ncentroids=args.ncentroids, |
| 253 | seed=args.seed, |
| 254 | code_size=args.code_size, |
| 255 | probe=args.probe |
| 256 | ) |
| 257 | |
| 258 | logger.info(f"Index saved to: {index_path}") |
| 259 | |
| 260 | if __name__ == "__main__": |
| 261 | main() |
no test coverage detected