Docs root

app.py

@@ -21,6 +21,9 @@ hf_token = os.getenv("HF_TOKEN", None)
 if hf_token is not None:
     login(token=hf_token)
 
+# Configurable model identifier
+model_name = os.getenv("HF_MODEL", "swiss-ai/Apertus-8B-Instruct-2509")
+
 # Keep data in session
 model = None
 tokenizer = None
@@ -42,9 +45,7 @@ async def lifespan(app: FastAPI):
     """Load the transformer model on startup"""
     global model, tokenizer
     try:
-        logger.info("Loading sentiment analysis model...")
-        # TODO: make this configurable
-        model_name = "swiss-ai/Apertus-8B-Instruct-2509"
+        logger.info(f"Loading model: {model_name}")
 
         # Automatically select device based on availability
         device = "cuda" if cuda.is_available() else "cpu"
@@ -73,6 +74,7 @@ app = FastAPI(
     title="Apertus API",
     description="REST API for serving Apertus models via Hugging Face transformers",
     version="0.1.0",
+    docs_url="/",
     lifespan=lifespan
 )
 
@@ -144,7 +146,5 @@ async def health_check():
         "gpu_available": cuda.is_available()
     }
 
-@app.get("/")
-def read_root():
-    return PlainTextResponse('Habemus Apertus')
-
+if __name__=='__main__':
+    uvicorn.run('app:app', reload=True)