test1

app.py

@@ -2,35 +2,42 @@ import gradio as gr
 from huggingface_hub import InferenceClient
 import os
 
-# Uses the Hugging Face token you just saved as a secret
-HF_TOKEN = os.getenv("KIMI_API_KEY") 
+# This pulls the 'HF_TOKEN' secret you just saved
+HF_TOKEN = os.getenv("HF_TOKEN")
 
-# We'll use a powerful open-source model hosted by Hugging Face
+# Using Llama-3-8B: Fast, smart, and runs on Hugging Face's own servers
 client = InferenceClient("meta-llama/Meta-Llama-3-8B-Instruct", token=HF_TOKEN)
 
 def samaran_kernel_chat(message, history):
-    system_message = "You are the Samaran Kernel. You are a privacy-first AI. Use deep reasoning and be witty."
+    # The Samaran Personality
+    system_message = "You are the Samaran Kernel. A privacy-focused AI collaborator. Be witty, insightful, and clear."
     
-    # Build the prompt
-    prompt = f"<|system|>\n{system_message}</s>\n"
+    # Format the conversation
+    messages = [{"role": "system", "content": system_message}]
     for user_msg, ai_msg in history:
-        prompt += f"<|user|>\n{user_msg}</s>\n<|assistant|>\n{ai_msg}</s>\n"
-    prompt += f"<|user|>\n{message}</s>\n<|assistant|>\n"
+        messages.append({"role": "user", "content": user_msg})
+        messages.append({"role": "assistant", "content": ai_msg})
+    messages.append({"role": "user", "content": message})
 
-    # Generate response
-    response = client.text_generation(prompt, max_new_tokens=512, stream=True)
+    # Stream the response
+    response = ""
+    for message_chunk in client.chat_completion(
+        messages,
+        max_tokens=512,
+        stream=True,
+    ):
+        token = message_chunk.choices[0].delta.content
+        if token:
+            response += token
+            yield response
 
-    partial_message = ""
-    for token in response:
-        partial_message += token
-        yield partial_message
-
-view = gr.ChatInterface(
+# The "T3-Style" User Interface
+demo = gr.ChatInterface(
     fn=samaran_kernel_chat,
-    title="Samaran Kernel (Privacy-First)",
-    description="Running on Open-Source Llama-3 via Hugging Face. No 3rd-party corporate data sharing.",
-    theme="soft"
+    title="Samaran Kernel",
+    description="Privacy-First AI Interface. Your data stays on Hugging Face.",
+    theme="glass" # Sleek, modern look
 )
 
 if __name__ == "__main__":
-    view.launch()
+    demo.launch()