missing tensor: backbone.vision_backbone.trunk.pos_embed

[likaigelikaige]

(FastSAM) PS I:\github\sam3cpp\build\Release> .\sam3-vision-trunk.exe .\sam3-image-f16.gguf "I:/github/OrtInference/assets/ocr/110167.jpg" "output" --cpu ggml_cuda_init: found 1 CUDA devices (Total VRAM: 8191 MiB): Device 0: Quadro P4000, compute capability 6.1, VMM: yes, VRAM: 8191 MiB missing tensor: backbone.vision_backbone.trunk.pos_embed