Update app.py

app.py

@@ -194,49 +194,43 @@ label_mapping = {
 
 #Function to classify user input
 def classifyTimeFrame(user_input):
-    while True:
-        #user_input = input("Enter a command (or type 'q' to quit): ")
-        if user_input.lower() == 'q':
-            print("Exiting...")
-            break
-             
-        # Tokenize and predict
-        input_encoding = tokenizer(user_input, padding=True, truncation=True, return_tensors="pt").to('cpu')
-        
-        with torch.no_grad():
-            attention_mask = input_encoding['attention_mask'].clone()
+    # Tokenize and predict
+    input_encoding = tokenizer(user_input, padding=True, truncation=True, return_tensors="pt").to('cpu')
+    
+    with torch.no_grad():
+        attention_mask = input_encoding['attention_mask'].clone()
 
 
 
-            # Modify the attention mask to emphasize certain key tokens
-            # for idx, token_id in enumerate(input_encoding['input_ids'][0]):
-            #     word = tokenizer.decode([token_id])
-            #     print(word)
-            #     if word.strip() in ["now", "same", "continue", "again", "also"]:  # Target key tokens
-            #         attention_mask[0, idx] = 3  # Increase attention weight for these words
-            #     else:
-            #         attention_mask[0, idx] = 0 
-            # print (attention_mask)
-            # input_encoding['attention_mask'] = attention_mask   
-            # print (input_encoding)
-            output = model(**input_encoding, output_hidden_states=True)
+        # Modify the attention mask to emphasize certain key tokens
+        # for idx, token_id in enumerate(input_encoding['input_ids'][0]):
+        #     word = tokenizer.decode([token_id])
+        #     print(word)
+        #     if word.strip() in ["now", "same", "continue", "again", "also"]:  # Target key tokens
+        #         attention_mask[0, idx] = 3  # Increase attention weight for these words
+        #     else:
+        #         attention_mask[0, idx] = 0 
+        # print (attention_mask)
+        # input_encoding['attention_mask'] = attention_mask   
+        # print (input_encoding)
+        output = model(**input_encoding, output_hidden_states=True)
 
-            probabilities = F.softmax(output.logits, dim=-1)
+        probabilities = F.softmax(output.logits, dim=-1)
 
-            prediction = torch.argmax(output.logits, dim=1).cpu().numpy()
+        prediction = torch.argmax(output.logits, dim=1).cpu().numpy()
 
-            # Map prediction back to label
-            print(prediction)
-            predicted_label = label_mapping[prediction[0]]
+        # Map prediction back to label
+        print(prediction)
+        predicted_label = label_mapping[prediction[0]]
 
 
-            print(f"Predicted intent: {predicted_label}\n")
-            # Print the confidence for each label
-            print("\nLabel Confidence Scores:")
-            for i, label in label_mapping.items():
-                confidence = probabilities[0][i].item()  # Get confidence score for each label
-                print(f"{label}: {confidence:.4f}")
-            print("\n")
+        print(f"Predicted intent: {predicted_label}\n")
+        # Print the confidence for each label
+        print("\nLabel Confidence Scores:")
+        for i, label in label_mapping.items():
+            confidence = probabilities[0][i].item()  # Get confidence score for each label
+            print(f"{label}: {confidence:.4f}")
+        print("\n")
 
 iface = gr.Interface(fn=classifyTimeFrame, inputs="text", outputs="text") 
 iface.launch(share=True)