Use upstream NequipTorchSimModel #400
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Member
Author
|
@cw-tan Can you give this a look over and okay? |
CompRhys
commented
Jan 9, 2026
| r_max: float, | ||
| type_names: list[str], | ||
| device: torch.device | None = None, | ||
| neighbor_list_fn: Callable = torchsim_nl, |
Member
Author
There was a problem hiding this comment.
Here this the main regression in so far as we loose the ability to vary the nl if using the nequip implementation upstream.
CompRhys
commented
Jan 9, 2026
| """Checkpoint download URLs for NequIP models.""" | ||
|
|
||
| Si = "https://github.com/abhijeetgangan/pt_model_checkpoints/raw/refs/heads/main/nequip/Si.nequip.pth" | ||
| # Cache directory for compiled models (under tests/ for easy cleanup) |
Member
Author
There was a problem hiding this comment.
Introduced just to speed up repeat testing locally. Nequip tests will take a long time in CI due to needing to compile step. Should consider if we can cache compiled models in CI also.
CompRhys
commented
Jan 9, 2026
CompRhys
commented
Jan 10, 2026
| assert model._device == DEVICE # noqa: SLF001 | ||
|
|
||
|
|
||
| # NOTE: we take [:-1] to skip benzene. This is because the stress calculation in NequIP |
Member
Author
There was a problem hiding this comment.
Benzene test failed on github CI runner. Similar numerical issue existed for SevenNet when PBC turned off. Stress is not meaningful when PBC is off.
CompRhys
commented
Jan 10, 2026
CompRhys
commented
Jan 10, 2026
Signed-off-by: Rhys Goodall <[email protected]>
Signed-off-by: Rhys Goodall <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Use the TorchSim Calc from https://github.com/mir-group/nequip rather than duplicating code. Not clear how to get dtype from AOTInductor, upstream to match the contract they just set the dtype to float64. Adds warning for end users. Implement basic cache for the compiled models to speed up repeat tests locally.