Update space

app.py

@@ -21,18 +21,18 @@ client = InferenceClient("microsoft/Phi-3-mini-4k-instruct")
 def respond(
     message,
     history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
+    # system_message,
+    # max_tokens,
+    # temperature,
+    # top_p,
 ):
-    messages = [{"role": "system", "content": system_message}]
+    messages = [{"role": "system", "content": "You are a helpful assistant who always talks like Yoda from Star Wars."}]
 
     for val in history:
         if val[0]:
             messages.append({"role": "user", "content": val[0]})
         if val[1]:
-            messages.append({"role": "Yoda from Star Wars.", "content": val[1]})
+            messages.append({"role": "assistant", "content": val[1]})
 
     messages.append({"role": "user", "content": message})
 
@@ -40,10 +40,10 @@ def respond(
 
     for message in client.chat_completion(
         messages,
-        max_tokens=max_tokens,
+        max_tokens=128,
         stream=True,
-        temperature=temperature,
-        top_p=top_p,
+        temperature=0.3,
+        top_p=50,
     ):
         token = message.choices[0].delta.content