generating lammps potential file using create_lammps_model.py

[michaelmacisaac]

Hi all,

I am wanting to use generate a LAMMPS model of one of the foundation models, but I am getting the error seen below. I have looked through the 'issues' and 'discussions', but couldnt find a suitable solution. If I missed it I apologize in advance.

Thanks for any and all help.

Command
python ~/mace/mace/cli/create_lammps_model.py mace-medium-density-agnesi-stress_cpu.model

Error:
python ~/mace/mace/cli/create_lammps_model.py mace-medium-density-agnesi-stress_cpu.model /blue/subhash/michaelmacisaac/.conda/envs/mace_env/lib/python3.10/site-packages/e3nn/o3/_wigner.py:10: FutureWarning: You are using torch.loadwithweights_only=False(the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value forweights_onlywill be flipped toTrue. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via torch.serialization.add_safe_globals. We recommend you start setting weights_only=Truefor any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. _Jd, _W3j_flat, _W3j_indices = torch.load(os.path.join(os.path.dirname(__file__), 'constants.pt')) /home/michaelmacisaac/mace/mace/cli/create_lammps_model.py:66: FutureWarning: You are usingtorch.loadwithweights_only=False(the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value forweights_onlywill be flipped toTrue. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via torch.serialization.add_safe_globals. We recommend you start setting weights_only=True for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. model = torch.load( Only one head found in the model: default. Skipping selection. Traceback (most recent call last): File "/home/michaelmacisaac/mace/mace/cli/create_lammps_model.py", line 92, in <module> main() File "/home/michaelmacisaac/mace/mace/cli/create_lammps_model.py", line 87, in main lammps_model_compiled = jit.compile(lammps_model) File "/blue/subhash/michaelmacisaac/.conda/envs/mace_env/lib/python3.10/site-packages/e3nn/util/jit.py", line 99, in compile compile( File "/blue/subhash/michaelmacisaac/.conda/envs/mace_env/lib/python3.10/site-packages/e3nn/util/jit.py", line 99, in compile compile( File "/blue/subhash/michaelmacisaac/.conda/envs/mace_env/lib/python3.10/site-packages/e3nn/util/jit.py", line 99, in compile compile( [Previous line repeated 1 more time] File "/blue/subhash/michaelmacisaac/.conda/envs/mace_env/lib/python3.10/site-packages/e3nn/util/jit.py", line 111, in compile mod = torch.jit.script(mod, **script_options) File "/blue/subhash/michaelmacisaac/.conda/envs/mace_env/lib/python3.10/site-packages/torch/jit/_script.py", line 1429, in script ret = _script_impl( File "/blue/subhash/michaelmacisaac/.conda/envs/mace_env/lib/python3.10/site-packages/torch/jit/_script.py", line 1147, in _script_impl return torch.jit._recursive.create_script_module( File "/blue/subhash/michaelmacisaac/.conda/envs/mace_env/lib/python3.10/site-packages/torch/jit/_recursive.py", line 557, in create_script_module return create_script_module_impl(nn_module, concrete_type, stubs_fn) File "/blue/subhash/michaelmacisaac/.conda/envs/mace_env/lib/python3.10/site-packages/torch/jit/_recursive.py", line 634, in create_script_module_impl create_methods_and_properties_from_stubs( File "/blue/subhash/michaelmacisaac/.conda/envs/mace_env/lib/python3.10/site-packages/torch/jit/_recursive.py", line 466, in create_methods_and_properties_from_stubs concrete_type._create_methods_and_properties( RuntimeError: Module 'reshape_irreps' has no attribute 'cueq_config' : File "/blue/subhash/michaelmacisaac/.conda/envs/mace_env/lib/python3.10/site-packages/mace/modules/irreps_tools.py", line 89 field = tensor[:, ix : ix + mul * d] # [batch, sample, mul * repr] ix += mul * d if hasattr(self, "cueq_config") and self.cueq_config is not None: ~~~~~~~~~~~~~~~~ <--- HERE if self.cueq_config.layout_str == "mul_ir": field = field.reshape(batch, mul, d)

[ilyes319]

Hello,

See that issue #741

it is the exact same problem, try the code I pasted.

[ilyes319]

Here is the code to use #741 (comment)

[yjw0510]

@ilyes319 I have the same problem. It works for MACE-OFF but it does work for loaded models

For example,

from mace.tools.scripts_utils import extract_config_mace_model
from mace.calculators import MACECalculator
from mace.modules import ScaleShiftMACE
import numpy as np
calc = MACECalculator(model_paths='/gpfs/yjw0510/mace_models/mace-mpa-0-medium.model',
                            device='cuda',
                            enable_cueq=False,
                            default_dtype="float32")
from mace.tools.scripts_utils import extract_config_mace_model

source_model = calc.models[0]
config = extract_config_mace_model(source_model)
target_model = source_model.__class__(**config).to("cuda")
target_model.load_state_dict(source_model.state_dict())
from e3nn.util import jit
script_model = jit.compile(target_model)

I get the following error. There seems to be a change in the assumed dictionary keys that cause this error.

Traceback (most recent call last):
  File "/gpfs/yjw0510/10_SolidMat/04_metadynamics/250105_opes_explore_lammps/changer_lammps.py", line 15, in <module>
    target_model.load_state_dict(source_model.state_dict())
  File "/user/yjw0510/python/lib/python3.12/site-packages/torch/nn/modules/module.py", line 2584, in load_state_dict
    raise RuntimeError(
RuntimeError: Error(s) in loading state_dict for ScaleShiftMACE:
        Unexpected key(s) in state_dict: "pair_repulsion_fn.r_max", "pair_repulsion_fn.cutoff.p", "pair_repulsion_fn.cutoff.r_max".

[ilyes319]

hey @yjw0510 what MACE version are you using. Make sure to use the latest main.