Update app.py

app.py

@@ -62,11 +62,11 @@ def compute_embeddings(selected_task, input_text):
     max_length = 2042
     processed_texts = [f'Instruct: {task_description}\nQuery: {input_text}']
     
-    batch_dict = self.tokenizer(processed_texts, max_length=max_length - 1, return_attention_mask=False, padding=False, truncation=True)
-    batch_dict['input_ids'] = [input_ids + [self.tokenizer.eos_token_id] for input_ids in batch_dict['input_ids']]
-    batch_dict = self.tokenizer.pad(batch_dict, padding=True, return_attention_mask=True, return_tensors='pt')
+    batch_dict = tokenizer(processed_texts, max_length=max_length - 1, return_attention_mask=False, padding=False, truncation=True)
+    batch_dict['input_ids'] = [input_ids + [tokenizer.eos_token_id] for input_ids in batch_dict['input_ids']]
+    batch_dict = tokenizer.pad(batch_dict, padding=True, return_attention_mask=True, return_tensors='pt')
     batch_dict = {k: v.to(device) for k, v in batch_dict.items()}
-    outputs = self.model(**batch_dict)
+    outputs = model(**batch_dict)
     embeddings = last_token_pool(outputs.last_hidden_state, batch_dict['attention_mask'])
     embeddings = F.normalize(embeddings, p=2, dim=1)
     embeddings_list = embeddings.detach().cpu().numpy().tolist()
@@ -80,10 +80,10 @@ def compute_similarity(selected_task, sentence1, sentence2, extra_sentence1, ext
         print(f"Selected task not found: {selected_task}")
         return f"Error: Task '{selected_task}' not found. Please select a valid task."
     # Compute embeddings for each sentence
-    embeddings1 = self.compute_embeddings(self.selected_task, sentence1)
-    embeddings2 = self.compute_embeddings(self.selected_task, sentence2)
-    embeddings3 = self.compute_embeddings(self.selected_task, extra_sentence1)
-    embeddings4 = self.compute_embeddings(self.selected_task, extra_sentence2)
+    embeddings1 = compute_embeddings(selected_task, sentence1)
+    embeddings2 = compute_embeddings(selected_task, sentence2)
+    embeddings3 = compute_embeddings(selected_task, extra_sentence1)
+    embeddings4 = compute_embeddings(selected_task, extra_sentence2)
     
     # Convert embeddings to tensors
     embeddings_tensor1 = torch.tensor(embeddings1).to(device).half()
@@ -92,9 +92,9 @@ def compute_similarity(selected_task, sentence1, sentence2, extra_sentence1, ext
     embeddings_tensor4 = torch.tensor(embeddings4).to(device).half()
   
     # Compute cosine similarity
-    similarity1 = self._compute_cosine_similarity(embeddings1, embeddings2)
-    similarity2 = self._compute_cosine_similarity(embeddings1, embeddings3)
-    similarity3 = self._compute_cosine_similarity(embeddings1, embeddings4)
+    similarity1 = compute_cosine_similarity(embeddings1, embeddings2)
+    similarity2 = compute_cosine_similarity(embeddings1, embeddings3)
+    similarity3 = compute_cosine_similarity(embeddings1, embeddings4)
 
     # Free memory
     free_memory(embeddings1, embeddings2, embeddings3, embeddings4)
@@ -102,7 +102,7 @@ def compute_similarity(selected_task, sentence1, sentence2, extra_sentence1, ext
     return similarity1, similarity2, similarity3
 
 # @spaces.GPU
-def _compute_cosine_similarity(emb1, emb2):
+def compute_cosine_similarity(emb1, emb2):
     tensor1 = torch.tensor(emb1).to(device).half()
     tensor2 = torch.tensor(emb2).to(device).half()
     similarity = F.cosine_similarity(tensor1, tensor2).item()