fairchem-core-1.3.0

What’s Changed

New Features / Enhancements

• EquiformerV2 + DeNS model and trainer (#880) @kyonofx
• Add wandb logger init to hydra runners (#894) @rayg1234
• replication use .item(); fix hydra_cli (#920) @kyonofx
• Add utils for logging weight tables and tensor stats (#907) @rayg1234
• Rename hydra heads (#903) @lbluque
• Updated loss and eval metrics (#896) @wood-b
• add hydra freeze backbone option (#898) @misko
• Make eqv2 old to hydra converter work with old old configs :'( (#856) @misko
• Read the logger logging level from environment variable (#874) @misko
• Add script to make release from fine tuned hydra (#875) @misko
• Add hydra entrypoint (#867) @rayg1234
• Add singleton logger (#873) @rayg1234
• add act checkpointing to escn (#852) @misko
• throw error when metadata is non integer (#865) @misko
• change logic to catch missing outputs (#859) @wood-b
• ESCN export II (#848) @rayg1234
• enable ruff formatter (#853) @misko

Bug Fixes

• add ema to BaseTrainer init (#916) @IliasChair
• Sort paths in datasets (#904) @lbluque
• remove destroy_process_group() from finally wrapper as it can hang (#884) @misko
• Add tests for hydra gemnet OC scaling factor generation and loading; Raise error on fail to load scaling factors (#831) @misko
• Match escn_exportable with escn main (#866) @misko
• fix bug where trainer state not being loaded (#863) @rayg1234

Dependencies

• update CI tests to py 3.12 (#858) @lbluque
• bump torch in pyproject.toml (#857) @lbluque

fairchem_data_oc-0.2.0

What’s Changed

Dependencies

• update CI tests to py 3.12 (#858) @lbluque

fairchem_applications_cattsunami-0.2.0

What’s Changed

Documentation

• adding tutorial (#882) @brookwander
• adding link to full dataset (#869) @brookwander

Dependencies

• bump torch in pyproject.toml (#857) @lbluque

fairchem_demo_ocpapi-0.2.0

What’s Changed

Dependencies

• ocpapi dependencies / requests -> 2.32.3 (#862) @kjmichel

fairchem_data_oc-0.1.0

What’s Changed

• Add solvent interface placement code (#765) @mshuaibii
• [BE] Remove large files from fairchem and add references to new location as needed (#761) @levineds
• [BE] Single ruff config (#751) @lbluque
• calc compatability with new configs (#724) @mshuaibii
• Package cleanup (#698) @lbluque
• PyPi CI workflows (#690) @lbluque
• cattsunami packaging (#678) @lbluque
• Fixes and updates to the documentation and website (#676) @zulissimeta
• Consolidate tests into a single folder (#682) @misko
• GA workflows update (#675) @lbluque
• Monorepo (#670) @misko

New Features / Enhancements

• add docstring. Add support for different pseudopotential path (#832) @mshuaibii

Bug Fixes

• explicitly define ispin for oc20 bulks (#840) @mshuaibii
• add tags, fix min_ab slab (#829) @mshuaibii

Deprecations

• Remove large file from repo prior to git history purge (#844) @levineds

fairchem-core-1.2.0

What’s Changed

New Features / Enhancements

• Do not shadow built in property (#841) @lbluque
• add script to port old equiv2 checkpoint+yaml to hydra version (#846) @misko
• Delete distributed option - Always use DDP (#833) @rayg1234
• Torch compile + export escn (#826) @rayg1234
• Add mean e to escn energy head (#828) @rayg1234
• add run group (#827) @rayg1234
• add option to apply mean to energy (#818) @rayg1234
• fix gemnet scaling factors fit.py and add a test (#819) @misko
• Add check to max num atoms (#817) @misko
• Use abs run_dir paths by default (#820) @rayg1234
• FM-v4 branch into main (#752) @misko
• refactor and deprecate old equiformerv2 (#812) @misko
• Fuse all hydras (#814) @rayg1234
• Rank2 tensor head (#792) @lbluque
• Explicitly initialize weights even if initialization method is "uniform" (#809) @misko
• Activation checkpoint equiformersv2 (#811) @rayg1234
• add resolution flag to escn (#804) @misko

Bug Fixes

• Add inference_only option to trainers (#850) @lbluque
• Sort dbs AseDBDataset (#837) @lbluque
• update to gemnet-oc hydra force head to work with amp (#825) @wood-b
• Load normalizers patch (#824) @lbluque
• load linref and normalizer on cuda device when resuming from checkpoint (#813) @misko

fairchem_applications_cattsunami-0.1.0

What’s Changed

• Fix NEB doc notebooks (#773) @brookwander
• [BE] Single ruff config (#751) @lbluque
• calc compatability with new configs (#724) @mshuaibii
• fixing issue with one reaction (#712) @brookwander
• adding seed to md tutorial and ipynb (#710) @brookwander
• Package cleanup (#698) @lbluque
• fixing summary fig and dissociation scheme fig in the gitbook (#700) @brookwander
• PyPi CI workflows (#690) @lbluque
• cattsunami packaging (#678) @lbluque

fairchem_demo_ocpapi-0.1.0

What’s Changed

• [BE] Single ruff config (#751) @lbluque
• no ASE version pin (#739) @lbluque
• move ocpapi to integration tests (#736) @rayg1234
• Package cleanup (#698) @lbluque
• PyPi CI workflows (#690) @lbluque
• Consolidate tests into a single folder (#682) @misko
• GA workflows update (#675) @lbluque
• fix ocpapi test (#681) @misko
• Monorepo (#670) @misko

Dependencies

• Bump requests from 2.31.0 to 2.32.0 in /packages/fairchem-demo-ocpapi (#707) @dependabot

fairchem_core-1.1.0

What’s Changed

• Allow symlink'd experimental folder by using absolute and not resolved paths (#798) @zulissimeta
• fix issues with ddp/hydra and add tests (#796) @misko
• Wrap atom coordinates in ase.Atoms preprocessing (#783) @kyonofx
• Add an option to run PBC in single system mode (#795) @misko
• Move select models to backbone + heads format and add support for hydra (#782) @misko
• [BE] Remove large files from fairchem and add references to new location as needed (#761) @levineds
• clone so3 embedding object (#781) @rayg1234
• add expandable segments var (#775) @rayg1234
• Fix dataset logic (#771) @lbluque
• Relax config (#758) @mshuaibii
• [BE] Single ruff config (#751) @lbluque
• Make relaxation data more general (#714) @anuroopsriram
• Add utils to help run torch profiling (#754) @rayg1234
• Updated ODAC checkpoints & configs (#755) @anuroopsriram
• Hide wandb.watch behind flag (#747) @rayg1234
• Try removing importlib - breaks multiprocess pickling (#746) @rayg1234
• No longer need to specify train dataset loader when predicting (#731) @curtischong
• Log num params to W&B (#741) @rayg1234
• Config fix (#737) @lbluque
• calc compatability with new configs (#724) @mshuaibii
• update preprocess script (#726) @mshuaibii
• support logger entity (#717) @mshuaibii
• pass seed to sampler (#697) @misko

New Features / Enhancements

• Finetune Hydra (#797) @rayg1234
• (OTF) Normalization and element references (#715) @lbluque
• Balanced batch sampler+base dataset (#753) @misko
• add proxy and ssl support. (#778) @haruyuki-oda
• Add slurm qos setting (#757) @rayg1234
• Equiformers2 Graph Parallel (#740) @rayg1234
• Add option to launch distributed runs locally with >1 GPU (#733) @rayg1234

Bug Fixes

• Fix job resume on non-preemption type failures (#803) @rayg1234
• #806: Move data.db to s3 and download via download_large_files (#807) @levineds

Documentation

• #806: Move data.db to s3 and download via download_large_files (#807) @levineds
• [BE] lint/test/docs on PR and push to main (#742) @lbluque

Dependencies

• pin ruff to 0.4.10 (#745) @rayg1234

fairchem_core-1.0.0

New repository structure!

We have restructured our codebase to have all of the FAIR Chemistry's code in one centralized repository of all its data, models, demos, and application efforts.

This release also includes releases for these other namespace packages

• fairchem_data_oc-0.0.1
• fairchem_demo_ocpapi-0.0.1
• fairchem_applications_cattsunami-0.0.1

What's Changed

New Features

• add unique key loader check to YAML to avoid problematic configs by @misko in #658
• [BE] PyPi integration + local model cache + Github actions by @misko in #623
• [BE] Add smoke test for [escn,gemnet,equiformer_v2] train+predict, Add optimization test for [escn,gemnet,equiformer_v2] by @misko in #640

Bug Fixes

• LBFGS batch size fix by @nimashoghi in #440

Other Changes

• Package scaling dictionary within GemNet-dT pretrained checkpoints by @abhshkdz in #419
• Updates bibtex entries for GemNet-OC, SCN, PaiNN by @abhshkdz in #420
• Update dimenet_plus_plus to use pyg activation function resolver by @zulissi in #427
• Use PBC configuration from ASE atoms objects by @emsunshine in #429
• Fixes GPU support with OCPCalculator (refs #424) by @abhshkdz in #430
• Use "deterministic" scatter on CUDA tensors if implemented by @abhshkdz in #438
• Support for using tag weights with L2MAE force loss by @abhshkdz in #439
• Zero out NaNs, if any, in the loss by @abhshkdz in #441
• Updating floating point precision for OC22/total_energy predictions and enabling OC22 challenge submission file generation by @wood-b in #421
• fix tag description by @mshuaibii in #444
• Upgrades to pytorch 1.13, cuda 11.6, pyg 2.2.0 by @abhshkdz in #442
• Relaxation storage optimizations by @mshuaibii in #445
• Support dataset sharding by @mshuaibii in #446
• Catch OOM during relaxations by @mshuaibii in #447
• Improper OOM merge fix by @mshuaibii in #451
• Warnings for missing scale files during validation by @zulissimeta in #450
• SCN checkpoint release + missing AdsorbML checkpoints by @mshuaibii in #453
• Adds scaling factors to GNOC-All-MD checkpoint and updates paths for some checkpoints by @abhshkdz in #456
• Updating flags and docs for training on total energy targets by @wood-b in #455
• eSCN checkpoints+configs by @mshuaibii in #457
• eSCN CPU support by @mshuaibii in #465
• LMDB folder fix by @zulissimeta in #476
• Revert "LMDB folder fix" by @abhshkdz in #480
• Allow models to include degenerate edges by @emsunshine in #467
• Adds init files to enable portability with pip install without the -e flag by @abhshkdz in #490
• Adds GemNet-OC checkpoint trained on OC20+OC22 total energies with enforce_max_neighbors_strictly=False by @abhshkdz in #495
• Fixes #501: don't cast a jagged list as np.array by @zulissimeta in #500
• Maintenance updates by @abhshkdz in #506
• clarify lmdb creation in tutorial by @brookwander in #509
• Add ASE Datasets by @emsunshine in #492
• LMDB-based ASE dataset, generic targets, target metadata guessing by @zulissimeta in #513
• Add type annotations by @r-barnes in #519
• Add more type annotations by @r-barnes in #522
• eSCN cpu fix by @mshuaibii in #523
• EquiformerV2 support by @abhshkdz in #518
• Add isort to CI by @r-barnes in #527
• ocp calc amp fix by @mshuaibii in #529
• Remove unused imports and sort imports by @r-barnes in #528
• Skipping scheduler setup in EquiformerV2 trainer if train_loader isn't defined by @abhshkdz in #541
• Add autoflake to pre-commit to eliminate unused imports automatically by @r-barnes in #534
• More type annotations by @r-barnes in #544
• Add more types, handle empty registries gracefully by @r-barnes in #545
• More type annotations by @r-barnes in #546
• Use logging instead of print by @r-barnes in #551
• More type annotations by @r-barnes in #548
• Bug Fixes for Tutorial by @emsunshine in #525
• Typecast scheduler params to int + fix slurm experiment logging by @abhshkdz in #555
• Adds a unit test for using OCP models with the ASE calculator by @abhshkdz in #565
• Fixes typo in loading checkpoints in BaseTrainer by @abhshkdz in #567
• Fixes Codecov badge URL by @abhshkdz in #568
• Adds boolean for whether to use strict max_neighbors in EquiformerV2 by @abhshkdz in #571
• GemNet-GP dense layer bugfix (fixes #573) by @abhshkdz in #574
• Adds an FAQ by @abhshkdz in #564
• Type fixes for using the MultistepLR scheduler by @abhshkdz in #578
• [Minor] Support for passing in stats to Equiformer V2 model by @anuroopsriram in #576
• Support IS2RE-Direct Training with ASE Read Datasets by @emsunshine in #579
• Fixes nested YAML config loading from the ASE calculator by @abhshkdz in #585
• Adds optional per-element linear reference coefficients to EquiformerV2 by @abhshkdz in #584
• Release EquiformerV2 checkpoint trained on OC22 by @abhshkdz in #586
• Changes for ODAC Release by @anuroopsriram in #588
• Update odac data link by @anuroopsriram in #595
• Update EqV2 31M ODAC checkpoint by @abhshkdz in #599
• Add natoms to ASE datasets by @emsunshine in #598
• DDEC release and IS2RS clarification for ODAC23 by @anuroopsriram in #602
• Unified OCP Trainer by @mshuaibii in #520
• Fix edge_index check in make_lmdb_sizes.py by @abhshkdz in #611
• cleanup+documentation by @mshuaibii in #615
• fix config spacing by @mshuaibii in #616
• Set default val and test dataset class to be same as train by @abhshkdz in #618
• bump miniconda version by @misko in #627
• Fix amp scale factor for loss by @janiceblue in #617
• Release ODAC23 data in extxyz format by @anuroopsriram in #625
• add seed option for calculator by @misko in #624
• Add in tests for coefficient mapping, mprimary, lprimary by @misko in #626
• Minor fix in run_relaxations by @AdeeshKolluru in #636
• Ase dataset updates by @lbluque in #622
• add step(batch) stateful sampler by @misko in #639
• Add warning and optimize sampler per @lbluque suggestion by @misko in #642
• lin ref in ase datasets by @lbluque in #643
• change _compute_metrics not to modify callers version by @misko in #645
• [BE] Set weights and biases as default logger by @janiceblue in #647
• Invert experimental import logic to opt-in by @levineds in #651
• do not add empty lines to include_dirs by @lbluque in https://...