Update to new HF endpoint + chat_completion

app.py

@@ -1,31 +1,35 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
+import os
 
-# Uncensored model ID
+# Your secret token
+token = os.getenv("HF_TOKEN")
+
+# Uncensored model
 model_id = "HuggingFaceH4/zephyr-7b-beta"
 
 def chat(message, history):
     try:
-        client = InferenceClient(model=model_id)
-        response = client.text_generation(
-            message,
-            max_new_tokens=200,
-            temperature=0.8,
-            do_sample=True,
-            stop_sequences=["User:", "\n\n"]
+        client = InferenceClient(token=token)
+        messages = [{"role": "user", "content": message}]
+        output = client.chat_completion(
+            messages,
+            model=model_id,
+            max_tokens=200,
+            temperature=0.8
         )
-        return response
+        return output.choices[0].message.content
     except Exception as e:
-        return f"Sorry, a quick glitch: {str(e)}. Try again!"
+        return f"Sorry, glitch: {str(e)}. Try again!"
 
-# Simple uncensored chat interface
+# Interface
 demo = gr.ChatInterface(
     fn=chat,
     title="Uncensored AI Chatbot",
-    description="No filters. Ask anything—dark jokes, opinions, anything!",
-    examples=["Tell me a dark, uncensored joke.", "Roast me like a pirate.", "Give a controversial opinion on [topic]."],
+    description="No filters. Ask anything.",
+    examples=["Tell me a dark, uncensored joke.", "Roast me like a pirate."],
     cache_examples=False
 )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch(auth=None)