using smaller size model binary

app.py

@@ -30,10 +30,10 @@ config.rope_theta = 10000.0
 model = LlamaForCausalLM(config)
 device = "cpu"
 model_id = "chbsaikiran/smollm2_135M_model"
-checkpoint_path = hf_hub_download(repo_id=model_id, filename="model_bin.pt")
+checkpoint_path = hf_hub_download(repo_id=model_id, filename="model_state_dict.pt")
 
 checkpoint = torch.load(checkpoint_path, map_location=device)
-model.load_state_dict(checkpoint['model_state_dict'])
+model.load_state_dict(checkpoint)
 model.to(device)
 model.eval()