snqsbqs

app.py

@@ -3,14 +3,7 @@ from huggingface_hub import InferenceClient
 
 client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
+def respond(message, history):
     if history is None:
         history = []
     
@@ -25,13 +18,12 @@ def respond(
     messages.append({"role": "user", "content": message})
     
     response = ""
-
     for message in client.chat_completion(
         messages,
-        max_tokens=max_tokens,
+        max_tokens=512,
         stream=True,
-        temperature=temperature,
-        top_p=top_p,
+        temperature=0.7,
+        top_p=0.95,
     ):
         token = message.choices[0].delta.content
         response += token
@@ -42,18 +34,23 @@ demo = gr.ChatInterface(
     additional_inputs=[], 
     examples=[
         ["How did the Leaning Tower of Pisa start leaning?"],
-        ["Who built the Colosseum?"],
-        ["What caused the fall of the Roman Empire?"],
-        ["When did World War II begin and end?"],
-        ["What was the significance of the Magna Carta?"],
         ["Comment la tour de Pise a-t-elle commencé à pencher ?"],
+        ["Who built the Colosseum?"],
         ["Qui a construit le Colisée ?"],
+        ["What caused the fall of the Roman Empire?"],
         ["Quelles sont les causes de la chute de l'Empire romain ?"],
+        ["When did World War II begin and end?"],
         ["Quand la Seconde Guerre mondiale a-t-elle commencé et terminé ?"],
+        ["What was the significance of the Magna Carta?"],
         ["Quelle est l'importance de la Magna Carta ?"]
     ],
-    cache_examples=False 
+    cache_examples=False,  
 )
 
+clear_button = gr.Button("Clear Chat")
+clear_button.click(lambda: None, None, demo.chatbot, queue=False)
+
+demo = gr.Column([demo, clear_button])
+
 if __name__ == "__main__":
     demo.launch()