First of all, congratulations on your amazing job. I am currently researching on estimating the 3D human pose and using this information to animate an avatar replicating the human's movements. To do it, I need to use lightweight software such as MotionAGFormer-XS or MotionAGFormer-S. However, when I use both of them, an error appears when running the code:
RuntimeError: Error(s) in loading state_dict for DataParallel:
Missing key(s) in state_dict: "module.layers.12.att_spatial.layer_scale_1", "module.layers.12.att_spatial.layer_scale_2", "module.layers.12.att_spatial.norm1.weight", "module.layers.12.att_spatial.norm1.bias", "module.layers.12.att_spatial.mixer.proj.weight", "module.layers.12.att_spatial.mixer.proj.bias", "module.layers.12.att_spatial.mixer.qkv.weight", "module.layers.12.att_spatial.norm2.weight", "module.layers.12.att_spatial.norm2.bias", "module.layers.12.att_spatial.mlp.fc1.weight", "module.layers.12.att_spatial.mlp.fc1.bias", "module.layers.12.att_spatial.mlp.fc2.weight", "module.layers.12.att_spatial.mlp.fc2.bias", "module.layers.12.att_temporal.layer_scale_1", "module.layers.12.att_temporal.layer_scale_2", "module.layers.12.att_temporal.norm1.weight", "module.layers.12.att_temporal.norm1.bias", "module.layers.12.att_temporal.mixer.proj.weight", "module.layers.12.att_temporal.mixer.proj.bias", "module.layers.12.att_temporal.mixer.qkv.weight", "module.layers.12.att_temporal.norm2.weight", "module.layers.12.att_temporal.norm2.bias", "module.layers.12.att_temporal.mlp.fc1.weight", "module.layers.12.att_temporal.mlp.fc1.bias", "module.layers.12.att_temporal.mlp.fc2.weight", "module.layers.12.att_temporal.mlp.fc2.bias", "module.layers.12.graph_spatial.layer_scale_1", "module.layers.12.graph_spatial.layer_scale_2", "module.layers.12.graph_spatial.norm1.weight", "module.layers.12.graph_spatial.norm1.bias", "module.layers.12.graph_spatial.mixer.U.weight", "module.layers.12.graph_spatial.mixer.U.bias", "module.layers.12.graph_spatial.mixer.V.weight", "module.layers.12.graph_spatial.mixer.V.bias", "module.layers.12.graph_spatial.mixer.batch_norm.weight", "module.layers.12.graph_spatial.mixer.batch_norm.bias", "module.layers.12.graph_spatial.mixer.batch_norm.running_mean", "module.layers.12.graph_spatial.mixer.batch_norm.running_var", "module.layers.12.graph_spatial.norm2.weight", "module.layers.12.graph_spatial.norm2.bias", "module.layers.12.graph_spatial.mlp.fc1.weight", "module.layers.12.graph_spatial.mlp.fc1.bias", "module.layers.12.graph_spatial.mlp.fc2.weight", "module.layers.12.graph_spatial.mlp.fc2.bias", "module.layers.12.graph_temporal.layer_scale_1", "module.layers.12.graph_temporal.layer_scale_2", "module.layers.12.graph_temporal.norm1.weight", "module.layers.12.graph_temporal.norm1.bias", "module.layers.12.graph_temporal.mixer.U.weight", "module.layers.12.graph_temporal.mixer.U.bias", "module.layers.12.graph_temporal.mixer.V.weight", "module.layers.12.graph_temporal.mixer.V.bias", "module.layers.12.graph_temporal.mixer.batch_norm.weight", "module.layers.12.graph_temporal.mixer.batch_norm.bias", "module.layers.12.graph_temporal.mixer.batch_norm.running_mean", "module.layers.12.graph_temporal.mixer.batch_norm.running_var", "module.layers.12.graph_temporal.norm2.weight", "module.layers.12.graph_temporal.norm2.bias", "module.layers.12.graph_temporal.mlp.fc1.weight", "module.layers.12.graph_temporal.mlp.fc1.bias", "module.layers.12.graph_temporal.mlp.fc2.weight", "module.layers.12.graph_temporal.mlp.fc2.bias", "module.layers.12.fusion.weight", "module.layers.12.fusion.bias", "module.layers.13.att_spatial.layer_scale_1", "module.layers.13.att_spatial.layer_scale_2", "module.layers.13.att_spatial.norm1.weight", "module.layers.13.att_spatial.norm1.bias", "module.layers.13.att_spatial.mixer.proj.weight", "module.layers.13.att_spatial.mixer.proj.bias", "module.layers.13.att_spatial.mixer.qkv.weight", "module.layers.13.att_spatial.norm2.weight", "module.layers.13.att_spatial.norm2.bias", "module.layers.13.att_spatial.mlp.fc1.weight", "module.layers.13.att_spatial.mlp.fc1.bias", "module.layers.13.att_spatial.mlp.fc2.weight", "module.layers.13.att_spatial.mlp.fc2.bias", "module.layers.13.att_temporal.layer_scale_1", "module.layers.13.att_temporal.layer_scale_2", "module.layers.13.att_temporal.norm1.weight", "module.layers.13.att_temporal.norm1.bias", "module.layers.13.att_temporal.mixer.proj.weight", "module.layers.13.att_temporal.mixer.proj.bias", "module.layers.13.att_temporal.mixer.qkv.weight", "module.layers.13.att_temporal.norm2.weight", "module.layers.13.att_temporal.norm2.bias", "module.layers.13.att_temporal.mlp.fc1.weight", "module.layers.13.att_temporal.mlp.fc1.bias", "module.layers.13.att_temporal.mlp.fc2.weight", "module.layers.13.att_temporal.mlp.fc2.bias", "module.layers.13.graph_spatial.layer_scale_1", "module.layers.13.graph_spatial.layer_scale_2", "module.layers.13.graph_spatial.norm1.weight", "module.layers.13.graph_spatial.norm1.bias", "module.layers.13.graph_spatial.mixer.U.weight", "module.layers.13.graph_spatial.mixer.U.bias", "module.layers.13.graph_spatial.mixer.V.weight", "module.layers.13.graph_spatial.mixer.V.bias", "module.layers.13.graph_spatial.mixer.batch_norm.weight", "module.layers.13.graph_spatial.mixer.batch_norm.bias", "module.layers.13.graph_spatial.mixer.batch_norm.running_mean", "module.layers.13.graph_spatial.mixer.batch_norm.running_var", "module.layers.13.graph_spatial.norm2.weight", "module.layers.13.graph_spatial.norm2.bias", "module.layers.13.graph_spatial.mlp.fc1.weight", "module.layers.13.graph_spatial.mlp.fc1.bias", "module.layers.13.graph_spatial.mlp.fc2.weight", "module.layers.13.graph_spatial.mlp.fc2.bias", "module.layers.13.graph_temporal.layer_scale_1", "module.layers.13.graph_temporal.layer_scale_2", "module.layers.13.graph_temporal.norm1.weight", "module.layers.13.graph_temporal.norm1.bias", "module.layers.13.graph_temporal.mixer.U.weight", "module.layers.13.graph_temporal.mixer.U.bias", "module.layers.13.graph_temporal.mixer.V.weight", "module.layers.13.graph_temporal.mixer.V.bias", "module.layers.13.graph_temporal.mixer.batch_norm.weight", "module.layers.13.graph_temporal.mixer.batch_norm.bias", "module.layers.13.graph_temporal.mixer.batch_norm.running_mean", "module.layers.13.graph_temporal.mixer.batch_norm.running_var", "module.layers.13.graph_temporal.norm2.weight", "module.layers.13.graph_temporal.norm2.bias", "module.layers.13.graph_temporal.mlp.fc1.weight", "module.layers.13.graph_temporal.mlp.fc1.bias", "module.layers.13.graph_temporal.mlp.fc2.weight", "module.layers.13.graph_temporal.mlp.fc2.bias", "module.layers.13.fusion.weight", "module.layers.13.fusion.bias", "module.layers.14.att_spatial.layer_scale_1", "module.layers.14.att_spatial.layer_scale_2", "module.layers.14.att_spatial.norm1.weight", "module.layers.14.att_spatial.norm1.bias", "module.layers.14.att_spatial.mixer.proj.weight", "module.layers.14.att_spatial.mixer.proj.bias", "module.layers.14.att_spatial.mixer.qkv.weight", "module.layers.14.att_spatial.norm2.weight", "module.layers.14.att_spatial.norm2.bias", "module.layers.14.att_spatial.mlp.fc1.weight", "module.layers.14.att_spatial.mlp.fc1.bias", "module.layers.14.att_spatial.mlp.fc2.weight", "module.layers.14.att_spatial.mlp.fc2.bias", "module.layers.14.att_temporal.layer_scale_1", "module.layers.14.att_temporal.layer_scale_2", "module.layers.14.att_temporal.norm1.weight", "module.layers.14.att_temporal.norm1.bias", "module.layers.14.att_temporal.mixer.proj.weight", "module.layers.14.att_temporal.mixer.proj.bias", "module.layers.14.att_temporal.mixer.qkv.weight", "module.layers.14.att_temporal.norm2.weight", "module.layers.14.att_temporal.norm2.bias", "module.layers.14.att_temporal.mlp.fc1.weight", "module.layers.14.att_temporal.mlp.fc1.bias", "module.layers.14.att_temporal.mlp.fc2.weight", "module.layers.14.att_temporal.mlp.fc2.bias", "module.layers.14.graph_spatial.layer_scale_1", "module.layers.14.graph_spatial.layer_scale_2", "module.layers.14.graph_spatial.norm1.weight", "module.layers.14.graph_spatial.norm1.bias", "module.layers.14.graph_spatial.mixer.U.weight", "module.layers.14.graph_spatial.mixer.U.bias", "module.layers.14.graph_spatial.mixer.V.weight", "module.layers.14.graph_spatial.mixer.V.bias", "module.layers.14.graph_spatial.mixer.batch_norm.weight", "module.layers.14.graph_spatial.mixer.batch_norm.bias", "module.layers.14.graph_spatial.mixer.batch_norm.running_mean", "module.layers.14.graph_spatial.mixer.batch_norm.running_var", "module.layers.14.graph_spatial.norm2.weight", "module.layers.14.graph_spatial.norm2.bias", "module.layers.14.graph_spatial.mlp.fc1.weight", "module.layers.14.graph_spatial.mlp.fc1.bias", "module.layers.14.graph_spatial.mlp.fc2.weight", "module.layers.14.graph_spatial.mlp.fc2.bias", "module.layers.14.graph_temporal.layer_scale_1", "module.layers.14.graph_temporal.layer_scale_2", "module.layers.14.graph_temporal.norm1.weight", "module.layers.14.graph_temporal.norm1.bias", "module.layers.14.graph_temporal.mixer.U.weight", "module.layers.14.graph_temporal.mixer.U.bias", "module.layers.14.graph_temporal.mixer.V.weight", "module.layers.14.graph_temporal.mixer.V.bias", "module.layers.14.graph_temporal.mixer.batch_norm.weight", "module.layers.14.graph_temporal.mixer.batch_norm.bias", "module.layers.14.graph_temporal.mixer.batch_norm.running_mean", "module.layers.14.graph_temporal.mixer.batch_norm.running_var", "module.layers.14.graph_temporal.norm2.weight", "module.layers.14.graph_temporal.norm2.bias", "module.layers.14.graph_temporal.mlp.fc1.weight", "module.layers.14.graph_temporal.mlp.fc1.bias", "module.layers.14.graph_temporal.mlp.fc2.weight", "module.layers.14.graph_temporal.mlp.fc2.bias", "module.layers.14.fusion.weight", "module.layers.14.fusion.bias", "module.layers.15.att_spatial.layer_scale_1", "module.layers.15.att_spatial.layer_scale_2", "module.layers.15.att_spatial.norm1.weight", "module.layers.15.att_spatial.norm1.bias", "module.layers.15.att_spatial.mixer.proj.weight", "module.layers.15.att_spatial.mixer.proj.bias", "module.layers.15.att_spatial.mixer.qkv.weight", "module.layers.15.att_spatial.norm2.weight", "module.layers.15.att_spatial.norm2.bias", "module.layers.15.att_spatial.mlp.fc1.weight", "module.layers.15.att_spatial.mlp.fc1.bias", "module.layers.15.att_spatial.mlp.fc2.weight", "module.layers.15.att_spatial.mlp.fc2.bias", "module.layers.15.att_temporal.layer_scale_1", "module.layers.15.att_temporal.layer_scale_2", "module.layers.15.att_temporal.norm1.weight", "module.layers.15.att_temporal.norm1.bias", "module.layers.15.att_temporal.mixer.proj.weight", "module.layers.15.att_temporal.mixer.proj.bias", "module.layers.15.att_temporal.mixer.qkv.weight", "module.layers.15.att_temporal.norm2.weight", "module.layers.15.att_temporal.norm2.bias", "module.layers.15.att_temporal.mlp.fc1.weight", "module.layers.15.att_temporal.mlp.fc1.bias", "module.layers.15.att_temporal.mlp.fc2.weight", "module.layers.15.att_temporal.mlp.fc2.bias", "module.layers.15.graph_spatial.layer_scale_1", "module.layers.15.graph_spatial.layer_scale_2", "module.layers.15.graph_spatial.norm1.weight", "module.layers.15.graph_spatial.norm1.bias", "module.layers.15.graph_spatial.mixer.U.weight", "module.layers.15.graph_spatial.mixer.U.bias", "module.layers.15.graph_spatial.mixer.V.weight", "module.layers.15.graph_spatial.mixer.V.bias", "module.layers.15.graph_spatial.mixer.batch_norm.weight", "module.layers.15.graph_spatial.mixer.batch_norm.bias", "module.layers.15.graph_spatial.mixer.batch_norm.running_mean", "module.layers.15.graph_spatial.mixer.batch_norm.running_var", "module.layers.15.graph_spatial.norm2.weight", "module.layers.15.graph_spatial.norm2.bias", "module.layers.15.graph_spatial.mlp.fc1.weight", "module.layers.15.graph_spatial.mlp.fc1.bias", "module.layers.15.graph_spatial.mlp.fc2.weight", "module.layers.15.graph_spatial.mlp.fc2.bias", "module.layers.15.graph_temporal.layer_scale_1", "module.layers.15.graph_temporal.layer_scale_2", "module.layers.15.graph_temporal.norm1.weight", "module.layers.15.graph_temporal.norm1.bias", "module.layers.15.graph_temporal.mixer.U.weight", "module.layers.15.graph_temporal.mixer.U.bias", "module.layers.15.graph_temporal.mixer.V.weight", "module.layers.15.graph_temporal.mixer.V.bias", "module.layers.15.graph_temporal.mixer.batch_norm.weight", "module.layers.15.graph_temporal.mixer.batch_norm.bias", "module.layers.15.graph_temporal.mixer.batch_norm.running_mean", "module.layers.15.graph_temporal.mixer.batch_norm.running_var", "module.layers.15.graph_temporal.norm2.weight", "module.layers.15.graph_temporal.norm2.bias", "module.layers.15.graph_temporal.mlp.fc1.weight", "module.layers.15.graph_temporal.mlp.fc1.bias", "module.layers.15.graph_temporal.mlp.fc2.weight", "module.layers.15.graph_temporal.mlp.fc2.bias", "module.layers.15.fusion.weight", "module.layers.15.fusion.bias".
It seems that the code when loading the model is not correlated with the weights. Do you have any idea? Also, just in case you changed the code, could you provide me with the code to load the model so that I can use those weights?
I appreciate any help you can provide.
Hello,
First of all, congratulations on your amazing job. I am currently researching on estimating the 3D human pose and using this information to animate an avatar replicating the human's movements. To do it, I need to use lightweight software such as MotionAGFormer-XS or MotionAGFormer-S. However, when I use both of them, an error appears when running the code:
It seems that the code when loading the model is not correlated with the weights. Do you have any idea? Also, just in case you changed the code, could you provide me with the code to load the model so that I can use those weights?
I appreciate any help you can provide.