[DPA-2] Is there any way to get the model or the output of fitting-net

[QuantumMisaka]

We have eval_descriptor method for DPA-2 model in Pytorch backend, so intuitively the fitting-net can be used to map the atomic descriptor to atomic energy and force, so :

• Is there any way to get the model parameters of fitting-net, or get the mapping of fitting-net from descriptor to atomic energy and forces ?
• Is there any way to quickly output the model architecture and the number of parameters in deepmd-kit ?

Thanks!

[njzjz]

• Is there any way to quickly output the model architecture and the number of parameters in deepmd-kit ?

dp convert-backend frozen_model.pth frozen_model.yaml

[QuantumMisaka]

Thanks for reply! but error occurs when I tried this:

(deepmd-3rc0) [mps@GPU819 v2-7-3h-100w]$ dp --pt convert-backend frozen_model.pth frozen_model.yaml
[2025-01-20 16:29:41,739] DEEPMD WARNING To get the best performance, it is recommended to adjust the number of threads by setting the environment variables OMP_NUM_THREADS, DP_INTRA_OP_PARALLELISM_THREADS, and DP_INTER_OP_PARALLELISM_THREADS. See https://deepmd.rtfd.io/parallelism/ for more information.
Traceback (most recent call last):
  File "/home/mps/miniconda3/envs/deepmd-3rc0/bin/dp", line 8, in <module>
    sys.exit(main())
             ^^^^^^
  File "/home/mps/miniconda3/envs/deepmd-3rc0/lib/python3.11/site-packages/deepmd/main.py", line 927, in main
    deepmd_main(args)
  File "/home/mps/miniconda3/envs/deepmd-3rc0/lib/python3.11/site-packages/deepmd/entrypoints/main.py", line 86, in main
    convert_backend(**dict_args)
  File "/home/mps/miniconda3/envs/deepmd-3rc0/lib/python3.11/site-packages/deepmd/entrypoints/convert_backend.py", line 27, in convert_backend
    out_hook(OUTPUT, data)
  File "/home/mps/miniconda3/envs/deepmd-3rc0/lib/python3.11/site-packages/deepmd/dpmodel/utils/serialization.py", line 108, in save_dp_model
    model_dict = traverse_model_dict(
                 ^^^^^^^^^^^^^^^^^^^^
  File "/home/mps/miniconda3/envs/deepmd-3rc0/lib/python3.11/site-packages/deepmd/dpmodel/utils/serialization.py", line 42, in traverse_model_dict
    model_obj[kk] = traverse_model_dict(
                    ^^^^^^^^^^^^^^^^^^^^
  File "/home/mps/miniconda3/envs/deepmd-3rc0/lib/python3.11/site-packages/deepmd/dpmodel/utils/serialization.py", line 42, in traverse_model_dict
    model_obj[kk] = traverse_model_dict(
                    ^^^^^^^^^^^^^^^^^^^^
  File "/home/mps/miniconda3/envs/deepmd-3rc0/lib/python3.11/site-packages/deepmd/dpmodel/utils/serialization.py", line 51, in traverse_model_dict
    model_obj = callback(model_obj)
                ^^^^^^^^^^^^^^^^^^^
  File "/home/mps/miniconda3/envs/deepmd-3rc0/lib/python3.11/site-packages/deepmd/dpmodel/utils/serialization.py", line 114, in <lambda>
    "dtype": x.dtype.name,
             ^^^^^^^
AttributeError: 'float' object has no attribute 'dtype'

Should I open an issue about this ?

[njzjz]

Fixed by #4565