Update app.py

app.py

@@ -57,7 +57,7 @@ def generate_text(prompt):
         top_k=40,
         repetition_penalty=1.1
         )
-    results = tokenizer.decode(outputs[0], clean_up_tokenization_spaces=False) #.replace(prompt,"")
+    results = tokenizer.decode(outputs[0], clean_up_tokenization_spaces=False)
     return results
 
 def merlin_chat(message, history):
@@ -119,7 +119,8 @@ def merlin_chat(message, history):
         #result = generate_text(prompt, model_path, parameters, headers)
         #result = model(prompt,return_full_text=False, max_new_tokens=256, temperature=0.8, repetition_penalty=1.1)
         #response = result[0]['generated_text']
-        response = generate_text(prompt)
+        result = generate_text(prompt)
+        response = result.replace(prompt,"")
         print(f"COMPLETION: {response}") # so we can see it in logs
         start = 0
         end = 0
@@ -131,7 +132,7 @@ def merlin_chat(message, history):
         if end<=0:
             continue
         cleanStr = cleanStr[:end]
-        messageStr = cleanStr + ']'
+        messageStr = prompt + cleanStr + ']'
         messages = json.loads(messageStr)
         message = messages[-1]
         if message['role'] != 'assistant':