|
Download DENIM_REPRODUCE.md from HaomingLuo/AgentFEM-Material-Loading-Memory: direct link, hf CLI and curl.
- Browser
- Download file 673 Bytes
-
https://huggingface.co/datasets/HaomingLuo/AgentFEM-Material-Loading-Memory/resolve/main/DENIM_REPRODUCE.md
- Command line
-
hf download hf://datasets/HaomingLuo/AgentFEM-Material-Loading-Memory/DENIM_REPRODUCE.md
-
curl -L -o DENIM_REPRODUCE.md https://huggingface.co/datasets/HaomingLuo/AgentFEM-Material-Loading-Memory/resolve/main/DENIM_REPRODUCE.md
673 Bytes
| # Reproduce DENIM closure v1 | |
| From the AgentFEM-Physics-Data project root with the configured `fenicsx-env`: | |
| ```bash | |
| python -m src.generate_t2_graybox_closure --count-per-family 16 --force | |
| python -m src.train_t2_graybox_closure | |
| python -m src.validate_t2_graybox_closure | |
| python -m src.plot_t2_graybox_closure | |
| python -m pytest -q tests/test_t2_graybox_discrete_energy.py tests/test_t2_graybox_closure.py | |
| ``` | |
| The data were generated with AgentFEM commit | |
| `e76f2a96f58cc12f47fd6fae996c22fdca300ed1`. Reproducing generation therefore | |
| requires AgentFEM and its FEniCSx environment. Loading the published DENIM | |
| weights only requires PyTorch and the included core implementation. | |