Update app.py

app.py

@@ -1,43 +1,35 @@
 import gradio as gr
-from openai import OpenAI
+from huggingface_hub import InferenceClient
 import os
 
-# The "Kernel" Configuration
-# Get your API key from Moonshot AI or OpenRouter
-API_KEY = os.getenv("KIMI_API_KEY") 
-BASE_URL = "https://api.moonshot.cn/v1" # Or your custom endpoint
+# Uses the Hugging Face token you just saved as a secret
+HF_TOKEN = os.getenv("KIMI_API_KEY") 
 
-client = OpenAI(api_key=API_KEY, base_url=BASE_URL)
+# We'll use a powerful open-source model hosted by Hugging Face
+client = InferenceClient("meta-llama/Meta-Llama-3-8B-Instruct", token=HF_TOKEN)
 
 def samaran_kernel_chat(message, history):
-    # 1. Prepare the Conversation Context
-    messages = [{"role": "system", "content": "You are the Samaran Kernel. Use <think> tags for deep reasoning. Be witty, technical, and precise."}]
+    system_message = "You are the Samaran Kernel. You are a privacy-first AI. Use deep reasoning and be witty."
+    
+    # Build the prompt
+    prompt = f"<|system|>\n{system_message}</s>\n"
     for user_msg, ai_msg in history:
-        messages.append({"role": "user", "content": user_msg})
-        messages.append({"role": "assistant", "content": ai_msg})
-    messages.append({"role": "user", "content": message})
+        prompt += f"<|user|>\n{user_msg}</s>\n<|assistant|>\n{ai_msg}</s>\n"
+    prompt += f"<|user|>\n{message}</s>\n<|assistant|>\n"
 
-    # 2. Call the Kimi K2 Engine
-    response = client.chat.completions.create(
-        model="moonshot-v1-32k", # Replace with k2-thinking if using private endpoint
-        messages=messages,
-        stream=True,
-        temperature=0.6 # Recommended for Kimi K2 stability
-    )
+    # Generate response
+    response = client.text_generation(prompt, max_new_tokens=512, stream=True)
 
     partial_message = ""
-    for chunk in response:
-        if chunk.choices[0].delta.content:
-            partial_message += chunk.choices[0].delta.content
-            yield partial_message
+    for token in response:
+        partial_message += token
+        yield partial_message
 
-# 3. The T3-Style UI Interface
 view = gr.ChatInterface(
     fn=samaran_kernel_chat,
-    title="Samaran Kernel T3 Chat",
-    description="Running on Kimi K2 Engine. Optimized for Deep Reasoning and Agentic Logic.",
-    theme="soft",
-    examples=["Explain the MoE architecture of Kimi K2.", "Draft a technical pitch for a new AI SaaS."]
+    title="Samaran Kernel (Privacy-First)",
+    description="Running on Open-Source Llama-3 via Hugging Face. No 3rd-party corporate data sharing.",
+    theme="soft"
 )
 
 if __name__ == "__main__":