An error occurred when using mace_create_lammps_model, and the cause could not be found.

[xiaomumu-fyl]

Traceback (most recent call last): File "/home/fangyl/anaconda3/envs/MACE/bin/mace_create_lammps_model", line 7, in sys.exit(main()) ^^^^^^ File "/home/fangyl/anaconda3/envs/MACE/lib/python3.11/site-packages/mace/cli/create_lammps_model.py", line 91, in main model = run_e3nn_to_cueq(copy.deepcopy(model)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/fangyl/anaconda3/envs/MACE/lib/python3.11/site-packages/mace/cli/convert_e3nn_cueq.py", line 213, in run transfer_weights( File "/home/fangyl/anaconda3/envs/MACE/lib/python3.11/site-packages/mace/cli/convert_e3nn_cueq.py", line 170, in transfer_weights target_model.load_state_dict(target_dict) File "/home/fangyl/anaconda3/envs/MACE/lib/python3.11/site-packages/torch/nn/modules/module.py", line 2624, in load_state_dict raise RuntimeError( RuntimeError: Error(s) in loading state_dict for ScaleShiftMACE: Unexpected key(s) in state_dict: "products.0.symmetric_contractions.weight", "products.1.symmetric_contractions.weight", "products.2.symmetric_contractions.weight".

[ilyes319]

hey @xiaomumu-fyl, what is your mace version and what model are you trying to convert?

[ilyes319]

Can you tell me what model is that? is it a custom trained one, a foundation model or a finetuned model?

[xiaomumu-fyl]

It is a train model

[ilyes319]

make sure you are not passing the -compiled model but the normal model.

[xiaomumu-fyl]

pip install cuequivariance cuequivariance-torch cuequivariance-ops-torch-cu12 --force-reinstall
After running this command, I solved this problem. Thank you for your concern！！！！
However, I have encountered a new problem after compiling lammps.

[xiaomumu-fyl]

If you have encountered similar problems, I would be very grateful if you could offer me some solutions.