Loren commited on
Commit
f2cc8b2
·
verified ·
1 Parent(s): 61e3813

Upload database.py

Browse files
Files changed (1) hide show
  1. database.py +8 -3
database.py CHANGED
@@ -21,6 +21,9 @@ from dotenv import load_dotenv
21
  import pyarrow as pa
22
  import pyarrow.compute as pc
23
 
 
 
 
24
  # Initialisations
25
  load_dotenv()
26
  HF_TOKEN = os.getenv('API_HF_TOKEN')
@@ -212,7 +215,8 @@ def fetch_query_results(query: str, k_model: int = 10,
212
  df["distance"] = df["faiss_id"].map(distance_map)
213
 
214
  if use_rerank:
215
- print("use_rerank = True")
 
216
  # Cross-encoder
217
  df["chunk_text"] = df["chunk_text"].str.replace(r'\s+', ' ', regex=True).str.strip()
218
  top_passages = df["chunk_text"].tolist()
@@ -226,9 +230,10 @@ def fetch_query_results(query: str, k_model: int = 10,
226
  # Garder top k_cross
227
  df_top = df.head(k_cross)
228
  else:
229
- print("use_rerank = False")
 
230
  df = df.sort_values(by="distance", ascending=False)
231
- df["cross_score"] = None
232
  # Garder top k_model
233
  df_top = df.head(k_model)
234
 
 
21
  import pyarrow as pa
22
  import pyarrow.compute as pc
23
 
24
+ import logging
25
+ logging.basicConfig(level=logging.DEBUG)
26
+
27
  # Initialisations
28
  load_dotenv()
29
  HF_TOKEN = os.getenv('API_HF_TOKEN')
 
215
  df["distance"] = df["faiss_id"].map(distance_map)
216
 
217
  if use_rerank:
218
+ print("use_rerank = True", flush=True)
219
+ logging.debug("use_rerank = True")
220
  # Cross-encoder
221
  df["chunk_text"] = df["chunk_text"].str.replace(r'\s+', ' ', regex=True).str.strip()
222
  top_passages = df["chunk_text"].tolist()
 
230
  # Garder top k_cross
231
  df_top = df.head(k_cross)
232
  else:
233
+ print("use_rerank = False", flush=True)
234
+ logging.debug("use_rerank = False")
235
  df = df.sort_values(by="distance", ascending=False)
236
+ df["cross_score"] = df["distance"]
237
  # Garder top k_model
238
  df_top = df.head(k_model)
239