Error loading DinoV3 state dict

Hello, I'm having trouble loading the DINOV3ImageEncoder model from the example implementation given here:
https://github.com/modelscope/DiffSynth-Studio/blob/main/examples/z_image/model_inference_low_vram/Z-Image-i2L.py

When I try to run the code verbatim, I get the following while loading the model. Any insights?

RuntimeError: Error(s) in loading state_dict for DINOv3ImageEncoder:
	Missing key(s) in state_dict: "model.layer.0.norm1.weight", "model.layer.0.norm1.bias", "model.layer.0.attention.k_proj.weight", "model.layer.0.attention.v_proj.weight", "model.layer.0.attention.q_proj.weight", "model.layer.0.attention.o_proj.weight", ...
	Unexpected key(s) in state_dict: "layer.0.attention.k_proj.weight", "layer.0.attention.o_proj.bias", "layer.0.attention.o_proj.weight", "layer.0.attention.q_proj.weight", "layer.0.attention.v_proj.weight", "layer.0.layer_scale1.lambda1", "layer.0.layer_scale2.lambda1", "layer.0.mlp.down_proj.bias", "layer.0.mlp.down_proj.weight", "layer.0.mlp.gate_proj.bias", "layer.0.mlp.gate_proj.weight", "layer.0.mlp.up_proj.bias", "layer.0.mlp.up_proj.weight", "layer.0.norm1.bias", "layer.0.norm1.weight", ...

Start with examples/z_image/model_inference_low_vram/Z-Image-i2L.py and trace how DINOv3ImageEncoder loads its state dict. Compare the expected keys beginning with "model." against the checkpoint keys beginning with "layer." and inspect the relevant model-loading entry point. Done means the example loads the encoder without missing or unexpected state-dict keys.