some debug messages

app.py

@@ -24,8 +24,10 @@ def predict(message, history):
 
     messages = "".join(["".join(["\n<human>:"+item[0], "\n<bot>:"+item[1]])
                 for item in history_transformer_format])
+    print("MESSAGES: ", messages)
 
     model_inputs = tokenizer([messages], return_tensors="pt") # .to("cuda")
+    print("MODEL INPUT:\n", model_inputs)
     streamer = TextIteratorStreamer(tokenizer, timeout=10., skip_prompt=True, skip_special_tokens=True)
     generate_kwargs = dict(
         model_inputs,