deepmodeling/deepmd-kit

View on GitHub

A deep learning package for many-body potential energy representation and molecular dynamics

Stars: 2k
Forks: 649
Open beginner issues: 0
Indexed issues: 162
Avg merge: 6d 18h
Merged PRs (30d): 15

Dominant language: Python
License: LGPL-3.0
Last GitHub push: Sep 14, 2026
Latest indexed: Sep 15, 2026
Contributing guide: Contributing guide
Code of conduct: No code of conduct
Beginner labels: good first issue

Open All

162 open issues indexed

Consider outputting serialization tree before training for unified display across backends Open

deepmodeling/deepmd-kit#5185 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
investigate why gr is None Open

branch: master lang: py todo comment type: todo

deepmodeling/deepmd-kit#5217 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Refactor Training Code Open

deepmodeling/deepmd-kit#5229 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Multi-task Shared Parameter Support via setattr Open

deepmodeling/deepmd-kit#5231 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Unified Data Processing for deepmd/npy Open

deepmodeling/deepmd-kit#5232 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Implement Modifier Support Open

deepmodeling/deepmd-kit#5236 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Automatic Format Conversion in dp using dpdata Open

deepmodeling/deepmd-kit#5237 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Benchmark Hessian Accuracy and Efficiency Open

deepmodeling/deepmd-kit#5239 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
unpin protobuf after TF is upgraded to 2.21 Open

branch: master lang: toml todo comment type: todo

deepmodeling/deepmd-kit#5274 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
tests: decouple API consistency coverage from numerical consistency coverage Open

enhancement

deepmodeling/deepmd-kit#5373 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[mixed_batch=True]: Replace SameNlocBatchSampler with Open

branch: master lang: py todo comment type: todo

deepmodeling/deepmd-kit#5386 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[Feature Request] Last layer feature (latent feature) output in deepmd-kit function or CLI Open

enhancement

deepmodeling/deepmd-kit#5396 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
(loading head_b's weights when head_a is requested, or Open

branch: master lang: py todo comment type: bug

deepmodeling/deepmd-kit#5429 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[BUG] Installation error : nvcc fatal : Unsupported gpu architecture 'compute_50' Open

bug

deepmodeling/deepmd-kit#5435 · 6 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
unpin protobuf after TF is upgraded to 2.21 Open

branch: master lang: toml todo comment type: todo

deepmodeling/deepmd-kit#5462 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Re-enable this in GitHub CUDA CI once the hosted/self-hosted Open

branch: master lang: py todo comment type: todo

deepmodeling/deepmd-kit#5473 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
SeZM's ``torch.compile`` / AOT-export code paths are only Open

branch: master lang: py todo comment type: todo

deepmodeling/deepmd-kit#5480 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
in either the FFN SO(3) grid or the post-aggregation SO(3) message is Open

branch: master lang: py todo comment type: bug

deepmodeling/deepmd-kit#5560 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Align data-stat sampling behavior between TF and PT backends Open

bug

deepmodeling/deepmd-kit#5594 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
perf: DPA4 .pt2 LAMMPS (C++ AOTI) inference 3-6x slower than Python (ASE) due to CPU-side edge construction and data marshalling Open

deepmodeling/deepmd-kit#5595 · 5 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[Code scan] Validate TensorFlow multi-device descriptor/neighbour dimensions before raw kernels Open

bug

deepmodeling/deepmd-kit#5623 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[Code scan] Ignore virtual atoms in dpmodel neighbor statistics Open

bug

deepmodeling/deepmd-kit#5630 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[Code scan] Close merge_lmdb source environments through the LMDB cache Open

bug

deepmodeling/deepmd-kit#5635 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[Code scan] Validate TensorFlow tabulation GPU sizes before launching kernels Open

bug

deepmodeling/deepmd-kit#5654 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[Code scan] Honor documented fparam/aparam shorthand in dpmodel DeepEval Open

bug

deepmodeling/deepmd-kit#5662 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[Code scan] Honor documented fparam/aparam shorthand in non-TF1 DeepEval backends Open

bug

deepmodeling/deepmd-kit#5666 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[Code scan] Freeze Paddle DPA3 parameters with stop_gradient Open

bug

deepmodeling/deepmd-kit#5686 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[Code scan] Symmetrize ASE calculator stress after reshaping virial Open

bug

deepmodeling/deepmd-kit#5689 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
for the total across systems we'd need to sum all matches; Open

branch: master lang: py todo comment type: todo

deepmodeling/deepmd-kit#5704 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
replace with dedicated DescriptorExtractor class after refactor. Open

branch: master lang: py todo comment type: todo

deepmodeling/deepmd-kit#5705 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[BUG] Source installation with PyTorch + CUDA backend fails due to undocumented build-system dependencies and incomplete CUDA path detection Open

bug

deepmodeling/deepmd-kit#5729 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[Feature Request] DPA-4 / PyTorch backend import path requires `e3nn`, but `e3nn` is only declared as an optional `dpa-adapt` dependency Open

bug enhancement

deepmodeling/deepmd-kit#5730 · 5 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[Feature Request] Improve pip installation path to handle CUDA Toolkit discovery for PyTorch/Inductor workflows Open

Docs enhancement

deepmodeling/deepmd-kit#5731 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[Feature Request] Implement DPLR dipole_charge modifier in dpmodel Open

enhancement

deepmodeling/deepmd-kit#5754 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
pt_expt: align backend support with argcheck schema Open

enhancement

deepmodeling/deepmd-kit#5755 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
jax: align training/model support with argcheck schema Open

enhancement

deepmodeling/deepmd-kit#5756 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
tf2: align backend support with argcheck schema Open

enhancement

deepmodeling/deepmd-kit#5757 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
TF backend loss: exclude mixed_type padding atoms (per-frame normalization) Open

bug

deepmodeling/deepmd-kit#5760 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
pt losses dens/population/denoise: not covered by mixed_type padding fix Open

bug

deepmodeling/deepmd-kit#5761 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[Feature Request] Source installation docs should pin supported PyTorch versions Open

Docs enhancement

deepmodeling/deepmd-kit#5766 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
fix(loss): normalize global RMSE display metrics per frame for mixed-size batches Open

bug

deepmodeling/deepmd-kit#5767 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Add HybridMuon optimizer support to JAX and TF2 for DPA4 training Open

enhancement

deepmodeling/deepmd-kit#5771 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
refactor: enforce canonical-graph invariants via factory-only construction (NeighborGraph / GraphTensorPack) Open

enhancement

deepmodeling/deepmd-kit#5777 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Move torch-only kernels under pt_expt Open

enhancement

deepmodeling/deepmd-kit#5782 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[pt_expt] NeighborGraph compiled training is energy-only; generalize the trace to non-energy fittings Open

enhancement

deepmodeling/deepmd-kit#5805 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[pt_expt] Graph-lower .pt2 export/inference is energy-only; non-energy models lack deployment consumers Open

enhancement

deepmodeling/deepmd-kit#5806 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
pt_expt: freeze-time opt-out for the with-comm AOTI artifact (halve .pt2 compile time for MP models) Open

enhancement

deepmodeling/deepmd-kit#5811 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
[BUG] pt-expt DPA1 graph lower fails with f_use_norm on NPY data Open

bug

deepmodeling/deepmd-kit#5813 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
feat(pt_expt): support epoch-based training horizons Open

enhancement

deepmodeling/deepmd-kit#5814 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
refactor(train): centralize stable gradient clipping and non-finite guards Open

enhancement

deepmodeling/deepmd-kit#5816 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026

deepmodeling/deepmd-kit

Get new issues in your inbox