Update app.py

app.py

@@ -8,10 +8,10 @@ print(os.environ)
 openai.api_base = os.environ.get("OPENAI_API_BASE")
 openai.api_key = os.environ.get("OPENAI_API_KEY")
 
-BASE_SYSTEM_MESSAGE = """"""
+BASE_SYSTEM_MESSAGE = """### System Prompt\n"""
 
 def make_prediction(prompt, max_tokens=None, temperature=None, top_p=None, top_k=None, repetition_penalty=None):
-    completion = openai.Completion.create(model="Open-Orca/OpenOrca-Platypus2-13B", prompt=prompt, max_tokens=max_tokens, temperature=temperature, top_p=top_p, top_k=top_k, repetition_penalty=repetition_penalty, stream=True, stop=["</s>", "<|im_end|>"])
+    completion = openai.Completion.create(model="Phind/Phind-CodeLlama-34B-v2", prompt=prompt, max_tokens=max_tokens, temperature=temperature, top_p=top_p, top_k=top_k, repetition_penalty=repetition_penalty, stream=True, stop=["</s>", "<|im_end|>"])
     for chunk in completion:
         yield chunk["choices"][0]["text"]
 
@@ -33,10 +33,10 @@ def chat(history, system_message, max_tokens, temperature, top_p, top_k, repetit
     history = history or []
 
     messages = BASE_SYSTEM_MESSAGE + system_message.strip() + "\n" + \
-               "\n".join(["\n".join(["User: "+item[0]+"<|end_of_turn|>", "Assistant: "+item[1]+"<|end_of_turn|>"])
+               "\n".join(["\n".join(["### User Message\n"+item[0]+"\n\n", "### Assistant\n"+item[1]+"\n\n"])
                           for item in history])
     # strip the last `<|end_of_turn|>` from the messages
-    messages = messages.rstrip("<|end_of_turn|>")
+    #messages = messages.rstrip("<|end_of_turn|>")
     # remove last space from assistant, some models output a ZWSP if you leave a space
     messages = messages.rstrip()