more

app.py

@@ -48,12 +48,20 @@ def respond(message, max_new_tokens, temperature, system_message="You are a help
                 temperature=temperature,
                 use_cache=True,
             )
-        print("[DEBUG] message:",messages)
-
+        promt = messages[0]['content']
+        promt += "assistant"
+        print("[DEBUG] prompt with assistant:",promt)
+        
         # Decode the generated tokens back to text
         generated_text = tokenizer.decode(output[0], skip_special_tokens=True)
         print("[DEBUG] Generated Text:", generated_text)
 
+        start_pos = generated_text.find(promt)
+        result_text = generated_text[start_pos + len(promt):]
+        print("[DEBUG] Result Text:", result_text)
+
+        #print("[DEBUG] Generated Text:", generated_text)
+
         # Clean up the response by removing unwanted parts (e.g., system and user info)
         cleaned_response = "".join(generated_text.split("\n")[9:])  # Assuming the response ends at the last line