Upload folder using huggingface_hub

app.py

@@ -1,18 +1,20 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
 
-# Use a model that supports text-generation
+# Use conversational endpoint
 client = InferenceClient("meta-llama/Llama-3.2-1B-Instruct")
 
 def generate(prompt, temperature=0.8, max_tokens=256):
     try:
-        response = client.text_generation(
-            prompt=prompt,
+        # Use conversational instead of text_generation
+        messages = [{"role": "user", "content": prompt}]
+        response = client.chat_completion(
+            messages=messages,
             temperature=temperature,
-            max_new_tokens=max_tokens,
+            max_tokens=max_tokens,
             do_sample=True if temperature > 0 else False
         )
-        return response
+        return response.choices[0].message.content
     except Exception as e:
         return f"Error: {str(e)}"