Update app.py

app.py

@@ -12,7 +12,7 @@ app = FastAPI()
 os.environ["HF_HOME"] = "/data/huggingface"
 
 # Use a CPU-compatible model (non-GPTQ)
-model_id = "meta-llama/Llama-3.2-3B-Instruct"
+model_id = "TinyLlama/TinyLlama-1.1B-Chat-v1.0"
 hf_token = os.environ.get("HF_TOKEN")
 
 # Load model and tokenizer (no GPU-specific args)