Update app.py

app.py

@@ -1,42 +1,16 @@
-# app.py
-
-# ── 0) Monkey-patch DNS for the HF Inference API hostname ─────────────────────
-import socket
-
-IP_POOL = [
-    "3.165.206.104",
-    "3.165.206.54",
-    "3.165.206.39",
-    "3.165.206.82",
-]
-_orig_getaddrinfo = socket.getaddrinfo
-
-def patched_getaddrinfo(host, port, family=0, socktype=0, proto=0, flags=0):
-    if host == "api-inference.huggingface.co":
-        return [
-            (family or socket.AF_INET,
-             socktype or socket.SOCK_STREAM,
-             proto  or socket.IPPROTO_TCP,
-             "",
-             (ip, port))
-            for ip in IP_POOL
-        ]
-    return _orig_getaddrinfo(host, port, family, socktype, proto, flags)
-
-socket.getaddrinfo = patched_getaddrinfo
-
-# ── 1) Imports & Client setup ─────────────────────────────────────────────────
-import os
 import gradio as gr
 from huggingface_hub import InferenceClient
+import socket
+try:
+    print("Resolving api-inference.huggingface.co...")
+    print(socket.gethostbyname("api-inference.huggingface.co"))
+except Exception as e:
+    print("DNS failed:", e)
 
-# Ensure you have set HF_TOKEN in your Space’s Secrets
-client = InferenceClient(
-    model="HuggingFaceH4/zephyr-7b-beta",
-    token=os.environ["HF_TOKEN"]
-)
+# 1) Initialize your HF client
+client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
-# ── 2) One-shot chat function ─────────────────────────────────────────────────
+# 2) A pure, one-shot function
 def respond(
     system_message: str,
     user_prompt: str,
@@ -44,6 +18,7 @@ def respond(
     temperature: float,
     top_p: float
 ) -> str:
+    
     messages = [
         {"role": "system", "content": system_message},
         {"role": "user",   "content": user_prompt}
@@ -53,19 +28,18 @@ def respond(
         max_tokens=max_tokens,
         temperature=temperature,
         top_p=top_p,
-        stream=False
+        stream=False    # <-- non-streaming
     )
+ 
+
+    
     return resp.choices[0].message.content.strip()
 
-# ── 3) Gradio Interface with Queue ────────────────────────────────────────────
+# 3) Create a non-streaming Interface
 demo = gr.Interface(
     fn=respond,
     inputs=[
-        gr.Textbox(
-            "You are a helpful and friendly assistant for children. "
-            "All responses must be safe and positive.",
-            label="System message"
-        ),
+        gr.Textbox("You are a helpful and friendly assistant for children.", label="System message"),
         gr.Textbox(placeholder="Type your question…", label="User prompt"),
         gr.Slider(1, 2048, 512, label="Max tokens"),
         gr.Slider(0.1, 4.0, 0.7, label="Temperature"),
@@ -76,8 +50,7 @@ demo = gr.Interface(
     flagging_mode="never",
 )
 
-# Enable the REST API on /api/queue/predict
-
-# ── 4) Launch ────────────────────────────────────────────────────────────────
+# 4) Enable the REST API and then launch
+demo.queue(api_open=True)
 if __name__ == "__main__":
     demo.launch()