Update app.py

app.py

@@ -21,7 +21,6 @@ def model_inference(
     input_dict, history, decoding_strategy, temperature, max_new_tokens,
     repetition_penalty, top_p, progress=gr.Progress(track_tqdm=True)
 ): 
-    yield "Processing ..."
     
     t0 = time.perf_counter()
     text = input_dict["text"]
@@ -79,7 +78,7 @@ def model_inference(
     thread.start()
     # thread.join()
 
-    yield "Thinking ..."
+    yield ""
 
     buffer = ""