Trying to convert dinov2 model

[jdp8]

Question

I tried to convert this model using the following command:

python -m scripts.convert --model_id nguyenkhoa/dinov2_Liveness_detection_v2.2.3 --quantize --task image-classification

but got the following error:

ValueError: Trying to export a dinov2 model, that is a custom or unsupported architecture, but no custom onnx configuration was passed as `custom_onnx_configs`. Please refer to https://huggingface.co/docs/optimum/main/en/exporters/onnx/usage_guides/export_a_model#custom-export-of-transformers-models for an example on how to export custom models. Please open an issue at https://github.com/huggingface/optimum/issues if you would like the model type dinov2 to be supported natively in the ONNX export.

I looked a bit into the custom_onnx_configs flag and found this conversion example. My question is regarding what should I pass to custom_onnx_configs for the conversion to work? I could pass gpt2 as used in the example but I'm wondering what is the correct custom_onnx_configs input for dinov2 models.

Thank you!

[xenova]

Hi there 👋 We did recently add support for dinov2 exports; can you try upgrade your version of Optimum to the latest version (1.24.0)?

[jdp8]

Hello! Thank you so much for your help, this allowed me to convert the model 💯