covert yolo model to buddymlir failed

[georgelin0825]

Describe the bug
I try to build a yolo model like example in examples resnet18 or lenet etc. when at the dynamo_compiler convert time there appear node name s1 and s2 which can't be convert make the build failed

To Reproduce

model = torch.hub.load("ultralytics/yolov5", "yolov5s", pretrained=True, trust_repo=True, skip_validation=True,autoshape=True) 
model = model.model.eval()
dynamo_compiler = DynamoCompiler(
    primary_registry=tosa.ops_registry,
    aot_autograd_decomposition=inductor_decomp,
)
data = torch.randn([1, 3, 640, 640])
# Import the model into MLIR module and parameters.
with torch.no_grad():
    graphs = dynamo_compiler.importer(model, data)

Screenshots
when I try to debug into the frontend.py It shows