BUG: Missing key(s) in state_dict: "detector.backbone.vision_backbone.trunk.pos_embed"

[quangkmhd]

I get error when inference with custom checkpoint. Train my custom data in rf100 with segment

return Sam3VideoPredictorMultiGPU( File "/home/qmask_quangnh58/detect/sam3/sam3/model/sam3_video_predictor.py", line 318, in init super().init(*model_args, **model_kwargs) File "/home/qmask_quangnh58/detect/sam3/sam3/model/sam3_video_predictor.py", line 43, in init build_sam3_video_model( File "/home/qmask_quangnh58/detect/sam3/sam3/model_builder.py", line 778, in build_sam3_video_model missing_keys, unexpected_keys = model.load_state_dict( File "/home/qmask_quangnh58/detect/sam3/sam3_venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 2629, in load_state_dict raise RuntimeError( RuntimeError: Error(s) in loading state_dict for Sam3VideoInferenceWithInstanceInteractivity: Missing key(s) in state_dict: "detector.backbone.vision_backbone.trunk.pos_embed", "detector.backbone.vision_backbone.trunk.patch_embed.proj.weight", "detector.backbone.vision_backbone.trunk.blocks.0.norm1.weight", "detector.backbone.vision_backbone.trunk.blocks.0.norm1.bias", "detector.backbone.vision_backbone.trunk.blocks.0.attn.freqs_cis", "detector.backbone.vision_backbone.trunk.blocks.0.attn.qkv.weight", "detector.backbone.vision_backbone.trunk.blocks.0.attn.qkv.bias", ............

[shojint]

Please refer to https://github.com/facebookresearch/sam3/issues/260