(self,
dataset,
node_dir_prefix_vl = None,
node_dir_prefix_text = None,
embed_model_name_vl = 'vidore/colqwen2-v1.0',
embed_model_name_text = 'BAAI/bge-m3',
topk=10,
gmm=False)
| 218 | class HybridSearchEngine: |
| 219 | |
| 220 | def __init__(self, |
| 221 | dataset, |
| 222 | node_dir_prefix_vl = None, |
| 223 | node_dir_prefix_text = None, |
| 224 | embed_model_name_vl = 'vidore/colqwen2-v1.0', |
| 225 | embed_model_name_text = 'BAAI/bge-m3', |
| 226 | topk=10, |
| 227 | gmm=False): |
| 228 | self.dataset = dataset |
| 229 | self.dataset_dir = os.path.join('./data', dataset) |
| 230 | self.img_dir = os.path.join(self.dataset_dir, 'img') |
| 231 | self.ppocr_dir = os.path.join(self.dataset_dir, 'bge_ingestion') |
| 232 | self.engine_vl = SearchEngine(dataset,node_dir_prefix=node_dir_prefix_vl,embed_model_name=embed_model_name_vl) |
| 233 | self.engine_text = SearchEngine(dataset,node_dir_prefix=node_dir_prefix_text,embed_model_name=embed_model_name_text) |
| 234 | self.topk = topk |
| 235 | self.gmm = gmm |
| 236 | |
| 237 | def search(self,query): |
| 238 | union_result = False |
nothing calls this directly
no test coverage detected