Update app.py

app.py

@@ -1,3 +1,4 @@
+import os
 from fastapi import FastAPI, Request, HTTPException, Depends
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.security import HTTPBearer, HTTPAuthorizationCredentials
@@ -8,7 +9,6 @@ import uvicorn
 app = FastAPI()
 security = HTTPBearer()
 
-# 1. CORS Configuration
 app.add_middleware(
     CORSMiddleware,
     allow_origins=["*"],
@@ -16,66 +16,45 @@ app.add_middleware(
     allow_headers=["*"],
 )
 
-# 2. RESTORED API KEY
 MY_API_KEY = "my-secret-key-456"
 
-# 3. LOAD MODEL (Q4 for Speed)
-print("--- Loading ChatMPT Engine (Q4) ---")
+# Load model without extra overhead
 llm = Llama(
     model_path="./model.gguf",
     n_ctx=2048,
-    n_threads=2, # Optimized for HF Free Tier
-    n_batch=128,
+    n_threads=2,
+    n_batch=512,
     verbose=False
 )
 
 def verify_token(credentials: HTTPAuthorizationCredentials = Depends(security)):
-    """Validates the Bearer token: my-secret-key-456"""
     if credentials.credentials != MY_API_KEY:
-        raise HTTPException(status_code=403, detail="Unauthorized: Invalid API Key")
+        raise HTTPException(status_code=403, detail="Unauthorized")
     return credentials.credentials
 
-@app.get("/")
-def health():
-    return {"status": "online", "brand": "ChatMPT"}
-
 @app.post("/v1/chat")
 async def chat(request: Request, _ = Depends(verify_token)):
     try:
         data = await request.json()
         user_input = data.get("prompt", "").strip()
 
-        if not user_input:
-            return JSONResponse(content={"reply": "I'm listening!"})
-
-        # SYSTEM PROMPT: Forces the AI to be ChatMPT and stay concise
-        system_msg = "You are ChatMPT, a helpful AI created by the ChatMPT Team. Be concise and direct."
-        prompt = f"<s>[INST] {system_msg}\n\n{user_input} [/INST]"
-
-        print(f"User: {user_input}")
+        # REMOVED the extra <s> here to fix your warning
+        prompt = f"[INST] {user_input} [/INST]"
 
-        # GENERATION SETTINGS
         output = llm(
             prompt,
-            max_tokens=300, 
+            max_tokens=512,
             temperature=0.7,
-            stop=["</s>", "[INST]", "User:"], 
+            stop=["[/INST]", "User:"],
             echo=False
         )
 
         reply = output["choices"][0]["text"].strip()
-        
-        # Identity Cleanup
-        reply = reply.replace("ChatGPT", "ChatMPT").replace("OpenAI", "ChatMPT Team")
-
-        if not reply:
-            reply = "I'm here! Could you please rephrase that?"
-
-        return JSONResponse(content={"reply": reply})
+        return JSONResponse(content={"reply": reply or "..."})
 
     except Exception as e:
         print(f"Error: {e}")
-        return JSONResponse(status_code=500, content={"reply": "Server Error. Try again."})
+        return JSONResponse(status_code=500, content={"reply": "SYSTEM_ERROR_RETRY"})
 
 if __name__ == "__main__":
     uvicorn.run(app, host="0.0.0.0", port=7860)