Switch to router.huggingface.co OpenAI-compatible chat API for LLM completions

app.py

@@ -1,15 +1,16 @@
+
 import gradio as gr
-from huggingface_hub import InferenceClient
-from huggingface_hub.utils import HfHubHTTPError
 import os
+import requests
 
 # Use Inference API with LLM only
 MODEL_ID = "meta-llama/Llama-3.2-3B-Instruct"
 token = os.getenv("HUGGING_FACE_HUB_TOKEN")
+ROUTER_URL = "https://router.huggingface.co/v1/chat/completions"
 
 def improve_description(user_input: str) -> str:
     """
-    Rewrites a rough project description using LLM.
+    Rewrites a rough project description using LLM via router.huggingface.co OpenAI-compatible API.
     """
     if not user_input or not user_input.strip():
         return "Please provide a description to improve."
@@ -21,31 +22,26 @@ def improve_description(user_input: str) -> str:
     if not token:
         return "Error: HUGGING_FACE_HUB_TOKEN environment variable is not set. Please configure your Hugging Face token."
 
+    headers = {
+        "Authorization": f"Bearer {token}",
+        "Content-Type": "application/json"
+    }
+    payload = {
+        "model": MODEL_ID,
+        "messages": [
+            {"role": "system", "content": "You are a technical writing assistant. Rewrite project descriptions to be clear, professional, and concise while preserving all technical details."},
+            {"role": "user", "content": f"Rewrite this project description professionally:\n\n{user_text}"}
+        ],
+        "max_tokens": 300,
+        "temperature": 0.3
+    }
     try:
-        # Initialize client with model and token - library handles endpoint automatically
-        client = InferenceClient(model=MODEL_ID, token=token)
-        messages = [
-            {
-                "role": "system",
-                "content": "You are a technical writing assistant. Rewrite project descriptions to be clear, professional, and concise while preserving all technical details."
-            },
-            {
-                "role": "user",
-                "content": f"Rewrite this project description professionally:\n\n{user_text}"
-            }
-        ]
-        
-        response = client.chat.completions.create(
-            messages=messages,
-            max_tokens=300,
-            temperature=0.3,
-        )
-        improved_text = response.choices[0].message.content.strip()
-        
+        resp = requests.post(ROUTER_URL, headers=headers, json=payload, timeout=60)
+        if resp.status_code != 200:
+            return f"Error: API request failed - {resp.status_code} {resp.reason}: {resp.text}"
+        data = resp.json()
+        improved_text = data["choices"][0]["message"]["content"].strip()
         return improved_text if improved_text else "Error: No response generated."
-            
-    except HfHubHTTPError as e:
-        return f"Error: API request failed - {str(e)}"
     except Exception as e:
         return f"Error: An unexpected error occurred - {str(e)}"