SirinootKK commited on
Commit
0375104
·
1 Parent(s): 3f8e821

fixed app.py

Browse files
Files changed (1) hide show
  1. app.py +8 -3
app.py CHANGED
@@ -184,15 +184,18 @@ class Chatbot:
184
  # Answer = self.model_pipeline(message, context)
185
  # return Answer
186
  def predict_semantic_search(self, message):
 
187
  message = message.strip()
188
- query_embedding = self.embedding_model.encode([message], convert_to_tensor=True)[0]
189
- corpus_embeddings = self.embedding_model.encode(self.df['Question'].tolist(), convert_to_tensor=True)
190
- hits = util.semantic_search(query_embedding.unsqueeze(0), corpus_embeddings, top_k=1)
191
  hit = hits[0][0]
192
  context = self.df['Context'][hit['corpus_id']]
193
  Answer = self.model_pipeline(message, context)
194
  return Answer
195
 
 
 
196
  def predict_without_faiss(self,message):
197
  MostSimilarContext = ""
198
  min_distance = 1000
@@ -212,6 +215,8 @@ class Chatbot:
212
  return Answer
213
 
214
  bot = ChatbotModel()
 
 
215
 
216
  """#Gradio"""
217
 
 
184
  # Answer = self.model_pipeline(message, context)
185
  # return Answer
186
  def predict_semantic_search(self, message):
187
+ corpus_embeddings = bot._chatbot.prepare_sentences_vector(bot._chatbot.get_embeddings(bot._chatbot.df['Context']))
188
  message = message.strip()
189
+ query_embedding = self.embedding_model.encode(message, convert_to_tensor=True)
190
+ query_embedding = query_embedding.to('cuda')
191
+ hits = util.semantic_search(query_embedding, corpus_embeddings, top_k=1)
192
  hit = hits[0][0]
193
  context = self.df['Context'][hit['corpus_id']]
194
  Answer = self.model_pipeline(message, context)
195
  return Answer
196
 
197
+
198
+
199
  def predict_without_faiss(self,message):
200
  MostSimilarContext = ""
201
  min_distance = 1000
 
215
  return Answer
216
 
217
  bot = ChatbotModel()
218
+ corpus_embeddings = bot._chatbot.get_embeddings(bot._chatbot.df['Context'])
219
+ corpus_embeddings = bot._chatbot.prepare_sentences_vector(corpus_embeddings)
220
 
221
  """#Gradio"""
222