Shape mismatch in load.state_dict()

[santurini]

I am trying to running demo.py on some personalized prompts but I am not able to load the model as I get the following error:

RuntimeError: Error(s) in loading state_dict for MovementConvEncoder:
	size mismatch for main.0.weight: copying a param with shape torch.Size([512, 259, 4]) from checkpoint, the shape in current model is torch.Size([512, 18, 4]).

And the command I am using is:

python demo.py --cfg ./configs/config_h3d_stage3.yaml --example ./demos/t2m.txt

[zsolt-csibi]

I have the same issue, with different dimension:

RuntimeError: Error(s) in loading state_dict for MovementConvEncoder:
        size mismatch for main.0.weight: copying a param with shape torch.Size([512, 259, 4]) from checkpoint, the shape in current model is torch.Size([512, 247, 4]).

[duan348733684]

I have the same issue, with different dimension:

RuntimeError: Error(s) in loading state_dict for MovementConvEncoder:
        size mismatch for main.0.weight: copying a param with shape torch.Size([512, 259, 4]) from checkpoint, the shape in current model is torch.Size([512, 247, 4]).

@zsolt-csibi have you solved this problem?

[duan348733684]

I am trying to running demo.py on some personalized prompts but I am not able to load the model as I get the following error:

RuntimeError: Error(s) in loading state_dict for MovementConvEncoder:
	size mismatch for main.0.weight: copying a param with shape torch.Size([512, 259, 4]) from checkpoint, the shape in current model is torch.Size([512, 18, 4]).

And the command I am using is:

python demo.py --cfg ./configs/config_h3d_stage3.yaml --example ./demos/t2m.txt

@santurini have you solved this problem?

[zrx-python]

Have you solved the dimension problem? Can you teach me the method @duan348733684

[zrx-python]

Have you solved the dimension problem? Can you teach me the method @zsolt-csibi

[ehsanwwe]

@zsolt-csibi @duan348733684 @santurini @zrx-python have you solve 🤔?

[zrx-python]

@ehsanwwe 解决了 但是忘记什么原因了，好像是某个超参数 还是 数据集不对的问题，你排查排查把 https://blog.csdn.net/qq_53930200/article/details/137102979?spm=1001.2014.3001.5501

[ehsanwwe]

( learning Chinese Lang is much important than learning Programming language 😄) @zrx-python i do all titles available at the link but everything's are the same

[SakuraHana01]

( learning Chinese Lang is much important than learning Programming language 😄) @zrx-python i do all titles available at the link but everything's are the same

@ehsanwwe have you solve this problem？

[Maxin-C]

This issue is likely related to the dataset version. I have explained the resolution process in #53, which might help address this problem. https://github.com/OpenMotionLab/MotionGPT/issues/53#issuecomment-2731960815