Improved output

Author: ChrisMaunder

multimode_llm.py

@@ -142,8 +142,6 @@ def do_chat(self, user_prompt: str, image: Image, system_prompt: str=None,
                     new_response = self.tokenizer_stream.decode(new_token)
                     response += new_response
 
-                    print(new_response)
-
                 inferenceMs = int((time.perf_counter() - start_inference_time) * 1000)
 
                 del generator

multimode_llm_adapter.py

@@ -105,8 +105,12 @@ def long_process(self, data: RequestData) -> JSON:
 
                     generator.compute_logits()
                     generator.generate_next_token()
-                    new_token = generator.get_next_tokens()[0]
-                    self.reply_text += tokenizer_stream.decode(new_token)
+
+                    next_tokens   = generator.get_next_tokens()
+                    next_token    = next_tokens[0]
+                    next_response = tokenizer_stream.decode(next_token)
+
+                    self.reply_text += next_response
 
             inferenceMs : int = int((time.perf_counter() - start_inference_time) * 1000)