I’m not able to upload my YOLOv11 weights on roboflow for auto label, I’m following the docs and using correct credentials still it is throwing this error :
This model upload failed. A failure usually occurs because of an older, incompatible model version or possibly a new type we don't currently have support for. Please try again with a new version and re-upload the model. If the error persists, reach out and we will be happy to help debug.
Hello @Ashish_IISC!
Thanks for reporting this, I’m Leandro, part of Roboflow team.
I took a look, and it seems the upload is failing during the YOLOv11 model conversion due to the following error:
"Error in main conversion logic: Error(s) in loading state_dict for DetectionModel: Missing key(s) in state_dict: "model.22.cv3.0.0.0.conv.weight", "model.22.cv3.0.0.0.bn.weight", "model.22.cv3.0.0.0.bn.bias", "model.22.cv3.0.0.0.bn.running_mean", "model.22.cv3.0.0.0.bn.running_var", "model.22.cv3.0.0.1.conv.weight", "model.22.cv3.0.0.1.bn.weight", "model.22.cv3.0.0.1.bn.bias", "model.22.cv3.0.0.1.bn.running_mean", "model.22.cv3.0.0.1.bn.running_var", "model.22.cv3.0.1.0.conv.weight", "model.22.cv3.0.1.0.bn.weight", "model.22.cv3.0.1.0.bn.bias", "model.22.cv3.0.1.0.bn.running_mean", "model.22.cv3.0.1.0.bn.running_var", "model.22.cv3.0.1.1.conv.weight", "model.22.cv3.0.1.1.bn.weight", "model.22.cv3.0.1.1.bn.bias", "model.22.cv3.0.1.1.bn.running_mean", "model.22.cv3.0.1.1.bn.running_var", "model.22.cv3.1.0.0.conv.weight", "model.22.cv3.1.0.0.bn.weight", "model.22.cv3.1.0.0.bn.bias", "model.22.cv3.1.0.0.bn.running_mean", "model.22.cv3.1.0.0.bn.running_var", "model.22.cv3.1.0.1.conv.weight", "model.22.cv3.1.0.1.bn.weight", "model.22.cv3.1.0.1.bn.bias", "model.22.cv3.1.0.1.bn.running_mean", "model.22.cv3.1.0.1.bn.running_var", "model.22.cv3.1.1.0.conv.weight", "model.22.cv3.1.1.0.bn.weight", "model.22.cv3.1.1.0.bn.bias", "model.22.cv3.1.1.0.bn.running_mean", "model.22.cv3.1.1.0.bn.running_var", "model.22.cv3.1.1.1.conv.weight", "model.22.cv3.1.1.1.bn.weight", "model.22.cv3.1.1.1.bn.bias", "model.22.cv3.1.1.1.bn.running_mean", "model.22.cv3.1.1.1.bn.running_var", "model.22.cv3.2.0.0.conv.weight", "model.22.cv3.2.0.0.bn.weight", "model.22.cv3.2.0.0.bn.bias", "model.22.cv3.2.0.0.bn.running_mean", "model.22.cv3.2.0.0.bn.running_var", "model.22.cv3.2.0.1.conv.weight", "model.22.cv3.2.0.1.bn.weight", "model.22.cv3.2.0.1.bn.bias", "model.22.cv3.2.0.1.bn.running_mean", "model.22.cv3.2.0.1.bn.running_var", "model.22.cv3.2.1.0.conv.weight", "model.22.cv3.2.1.0.bn.weight", "model.22.cv3.2.1.0.bn.bias", "model.22.cv3.2.1.0.bn.running_mean", "model.22.cv3.2.1.0.bn.running_var", "model.22.cv3.2.1.1.conv.weight", "model.22.cv3.2.1.1.bn.weight", "model.22.cv3.2.1.1.bn.bias", "model.22.cv3.2.1.1.bn.running_mean", "model.22.cv3.2.1.1.bn.running_var". Unexpected key(s) in state_dict: "model.22.cv3.0.0.conv.weight", "model.22.cv3.0.0.bn.weight", "model.22.cv3.0.0.bn.bias", "model.22.cv3.0.0.bn.running_mean", "model.22.cv3.0.0.bn.running_var", "model.22.cv3.0.0.bn.num_batches_tracked", "model.22.cv3.0.1.conv.weight", "model.22.cv3.0.1.bn.weight", "model.22.cv3.0.1.bn.bias", "model.22.cv3.0.1.bn.running_mean", "model.22.cv3.0.1.bn.running_var", "model.22.cv3.0.1.bn.num_batches_tracked", "model.22.cv3.1.0.conv.weight", "model.22.cv3.1.0.bn.weight", "model.22.cv3.1.0.bn.bias", "model.22.cv3.1.0.bn.running_mean", "model.22.cv3.1.0.bn.running_var", "model.22.cv3.1.0.bn.num_batches_tracked", "model.22.cv3.1.1.conv.weight", "model.22.cv3.1.1.bn.weight", "model.22.cv3.1.1.bn.bias", "model.22.cv3.1.1.bn.running_mean", "model.22.cv3.1.1.bn.running_var", "model.22.cv3.1.1.bn.num_batches_tracked", "model.22.cv3.2.0.conv.weight", "model.22.cv3.2.0.bn.weight", "model.22.cv3.2.0.bn.bias", "model.22.cv3.2.0.bn.running_mean", "model.22.cv3.2.0.bn.running_var", "model.22.cv3.2.0.bn.num_batches_tracked", "model.22.cv3.2.1.conv.weight", "model.22.cv3.2.1.bn.weight", "model.22.cv3.2.1.bn.bias", "model.22.cv3.2.1.bn.running_mean", "model.22.cv3.2.1.bn.running_var", "model.22.cv3.2.1.bn.num_batches_tracked". "
This usually indicates that the model was either trained or exported using an incompatible version of the Ultralytics framework. From the structure of the weights, it actually looks more like a YOLOv8 model rather than a YOLOv11 one.
I also noticed you were able to successfully upload a versionless YOLOv8 model earlier. However, at the moment, Auto Label does not support versionless models — that’s likely why you’re running into issues now.
To fix this and enable Auto Label with your model, I recommend creating a new dataset version and uploading the same model weights again, but this time using the -v <version> flag in the CLI to specify the dataset version explicitly.
You can find more details on how to upload versioned models here:
Upload Custom Weights | Roboflow Docs
Let us know if you hit any issues — happy to help further!