코퍼스 1287 (PDF 481) · 질의 752 {'A': 150, 'A_': 133, 'B': 149, 'C': 150, 'D': 150, 'E': 20}
C:\Users\USER\Desktop\DS_development\civil_agent.wt\symbol-retrieval-exp\python\civil_engine\rag\embed.py:28: UserWarning: The model sentence-transformers/paraphrase-multilingual-mpnet-base-v2 now uses mean pooling instead of CLS embedding. In order to preserve the previous behaviour, consider either pinning fastembed version to 0.5.1 or using `add_custom_model` functionality.
  return TextEmbedding(model_name=_MODEL_NAME)
  임베딩 doc_raw: 1287개 246s
  임베딩 doc_dec: 1287개 254s
  임베딩 q_752: 752개 10s
복합명사 사전 1800개 (예: ['가능하다면', '가능하도록', '가능해야', '가동받침', '가동받침의', '가동지점에서는', '가로방향', '가로흔들이가'])
V0    12s  hybrid Hit@3 A=0.033  A_=0.000  B=0.148  C=0.367  D=0.267  E=0.650
H1    12s  hybrid Hit@3 A=0.560  A_=0.068  B=0.483  C=0.333  D=0.313  E=0.650
H2    12s  hybrid Hit@3 A=0.627  A_=0.376  B=0.577  C=0.327  D=0.307  E=0.650
H3    12s  hybrid Hit@3 A=0.627  A_=0.376  B=0.564  C=0.340  D=0.307  E=0.650
H4    11s  hybrid Hit@3 A=0.640  A_=0.346  B=0.577  C=0.567  D=0.433  E=0.650
H5    10s  hybrid Hit@3 A=0.707  A_=0.398  B=0.584  C=0.560  D=0.433  E=0.600