What's Changed
Highlights
DPA3
DPA3 is an advanced interatomic potential leveraging the message-passing architecture. Designed as a large atomic model (LAM), DPA3 is tailored to integrate and simultaneously train on datasets from various disciplines, encompassing diverse chemical and materials systems across different research domains. Its model design ensures exceptional fitting accuracy and robust generalization within and beyond the training domain. Furthermore, DPA3 maintains energy conservation and respects the physical symmetries of the potential energy surface, making it a dependable tool for a wide range of scientific applications.
Refer to examples/water/dpa3/input_torch.json for the training script. After training, the PyTorch model can be converted to the JAX model.
PaddlePaddle backend
The PaddlePaddle backend features a similar Python interface to the PyTorch backend, ensuring compatibility and flexibility in model development. PaddlePaddle has introduced dynamic-to-static functionality and PaddlePaddle JIT compiler (CINN) in DeePMD-kit, which allow for dynamic shapes and higher-order differentiation. The dynamic-to-static functionality automatically captures the user’s dynamic graph code and converts it into a static graph. After conversion, the CINN compiler is used to optimize the computational graph, thereby enhancing the efficiency of model training and inference. In experiments with the DPA-2 model, we achieved approximately a 40% reduction in training time compared to the dynamic graph, effectively improving the model training efficiency.
Breaking changes
Other new features
- feat(pt/dp): support case embedding and sharable fitting by @iProzd in #4417
- feat(pt): train with energy Hessian by @1azyking in #4169
- feat: add new batch size rules for large systems by @caic99 in #4659
- feat: add method to access fele in pppm/dplr by @HanswithCMY in #4452
- feat (tf/pt): add atomic weights to tensor loss by @ChiahsinChu in #4466
- feat(pt): add
trainableto property fitting by @ChiahsinChu in #4599 - Feat(pt): Support fitting_net input statistics. by @Chengqian-Zhang in #4504
- feat(jax): Hessian by @njzjz in #4649
- feat: add plugin mode for data modifier by @ChiahsinChu in #4621
- feat(pt): add eta message for pt backend by @HydrogenSulfate in #4725
- feat: add huber loss by @iProzd in #4684
- feat(pt): add AdamW for pt training by @iProzd in #4757
- Feat:support customized rglob by @anyangml in #4763
- feat(pt/pd): add size option to dp show by @iProzd in #4783
All changes in v3.0.1, v3.0.2, and v3.0.3 are included.
Contributors
- @iProzd #4417 #4655 #4419 #4609 #4633 #4647 #4675 #4684 #4730 #4757 #4754 #4756 #4760 #4778 #4781 #4783 #4792
- @pre-commit-ci #4420 #4449 #4464 #4473 #4497 #4521 #4539 #4552 #4566 #4574 #4579 #4596 #4602 #4611 #4645 #4660 #4672 #4690 #4699 #4708 #4712 #4719 #4723 #4736 #4748 #4767 #4779 #4791
- @njzjz #4482 #4483 #4484 #4507 #4619 #4410 #4438 #4442 #4446 #4459 #4485 #4479 #4508 #4534 #4531 #4542 #4550 #4553 #4557 #4561 #4565 #4570 #4575 #4547 #4582 #4613 #4624 #4558 #4638 #4636 #4640 #4649 #4668 #4680 #4720 #4728 #4738 #4692 #4700 #4704 #4702 #4717 #4724 #4726 #4729 #4735 #4753 #4774 #4765 #4776 #4775 #4766 #4780 #4786 #4794
- @Chengqian-Zhang #4471 #4504 #4639
- @HydrogenSulfate #4418 #4489 #4673 #4302 #4439 #4414 #4480 #4493 #4488 #4512 #4467 #4514 #4617 #4556 #4656 #4694 #4701 #4715 #4725 #4768 #4770
- @QuantumMisaka #4510
- @1azyking #4169
- @caic99 #4535 #4615 #4659 #4434 #4426 #4435 #4433 #4437 #4463 #4505 #4478 #4541 #4513 #4597 #4622 #4662 #4669 #4677 #4678 #4688 #4687 #4737 #4747 #4746 #4761 #4772 #4773 #4784 #4751 #4790
- @dependabot #4408 #4630
- @anyangml #4423 #4432 #4587 #4763
- @HanswithCMY #4452
- @ChiahsinChu #4466 #4538 #4599 #4621
- @RMeli #4577
- @Yi-FanLi #4581
- @wanghan-iapcm #4653
- @SumGuo-88 #4593
- @SigureMo #4664
- @njzjz-bot #4796
New Contributors
- @HanswithCMY made their first contribution in #4452
- @QuantumMisaka made their first contribution in #4510
- @1azyking made their first contribution in #4169
- @RMeli made their first contribution in #4577
- @SumGuo-88 made their first contribution in #4593
- @SigureMo made their first contribution in #4664
Full Changelog: v3.0.0...v3.1.0rc0
