> 19 June 2023
- Add WBM element heatmap to site and data/wbm/readme.md [`6`](https://github.com/janosh/matbench-discovery/pull/6)
- doc: remove reference to alignn [`4`](https://github.com/janosh/matbench-discovery/pull/4)
- Initial PyPI release preparation [`5`](https://github.com/janosh/matbench-discovery/pull/5)
- Start work on website to host docs and display interactive leaderboard [`3`](https://github.com/janosh/matbench-discovery/pull/3)
- move m3gnet formation energy corrections to join_m3gnet_results.py [`248a79b`](https://github.com/janosh/matbench-discovery/commit/248a79bd755917c9e2f9a00c0f70c3e927321743)
- generate MEGNet formation energy preds for M3GNet-relaxed structures in join_m3gnet_results.py [`e526e62`](https://github.com/janosh/matbench-discovery/commit/e526e627119b537521e5f38d4b34ff28082245d8)
- merge load_model_preds() into load_df_wbm_with_preds() and pass through **kwargs to pandas reader [`1681c38`](https://github.com/janosh/matbench-discovery/commit/1681c38ed21afd119bd271eb9f14a67f6bf1a17c)
- add SI discussion on 'WBM Batch Robustness as a Measure of Extrapolation Prowess' [`705e0d0`](https://github.com/janosh/matbench-discovery/commit/705e0d0c1dc9ff6a17158bae4723eef9d031329d)
- add descriptions to all model metadata.yml [`4013bb1`](https://github.com/janosh/matbench-discovery/commit/4013bb18fd7bf36dcb4765e65d16f7838ff8144d)
- make metrics-table sortable with svelte-zoo action [`13746f4`](https://github.com/janosh/matbench-discovery/commit/13746f40a37b8d03131d2c942398b1141f0a35e9)
- initial commit for paper [`cd4b90f`](https://github.com/janosh/matbench-discovery/commit/cd4b90f9df529b382e47ab623e95cc92c2ea2ce6)
- add stable_metrics() to matbench_discovery/energy.py (formerly part of hist_classified_stable_vs_hull_dist() which gets a big refactor) [`f0363b4`](https://github.com/janosh/matbench-discovery/commit/f0363b4d2bd2e56cfc2db5aa48632ec1b3372611)
- fix WBM e_above_convex_hull values w.r.t. MP PPD [`5dae951`](https://github.com/janosh/matbench-discovery/commit/5dae951de222ccd6e0e6d0d41b6570e4ac1e0d04)
- add matbench_discovery/metrics.py to centralize computing metrics for plotting scripts [`500e670`](https://github.com/janosh/matbench-discovery/commit/500e67060be6e252d77eecae799a9c4032873152)
- add 2023-01-25-rolling-mae-vs-hull-dist-models.svelte [`9fed210`](https://github.com/janosh/matbench-discovery/commit/9fed210cf441e06f62c10ca1db4ae0a64a26b55b)
- add data/wbm/fetch_process_wbm_dataset.py [`6ac75aa`](https://github.com/janosh/matbench-discovery/commit/6ac75aa1c860eb28e6ad806a54c62d0a778aaac9)
- extract ModelCard component from page /models [`f5c3e37`](https://github.com/janosh/matbench-discovery/commit/f5c3e3775f96ffb3835aafe4f2f0a7de4eba9c04)
- add test_hist_classified_stable_as_func_of_hull_dist() [`c855e51`](https://github.com/janosh/matbench-discovery/commit/c855e512bc27a69ad03ed5de0223ba79fc5a49e7)
- add site/src/figs/2023-02-01-mp-elemental-ref-energies.svelte generated in data/wbm/analysis.py [`9f43711`](https://github.com/janosh/matbench-discovery/commit/9f43711c06fcd3d1c9fa018b6ed2835c8e2df8f3)
- allow multiple models per metadata.yml file, add models M3GNet + MEGNet and CGCNN+P to respective model cards [`24f6868`](https://github.com/janosh/matbench-discovery/commit/24f68688f68b6bb4561b03c8d9869074abd3bae5)
- refactor cumulative_clf_metric()->cumulative_precision_recall() to plot both prec and recall for all models [`4972c01`](https://github.com/janosh/matbench-discovery/commit/4972c010523f4fc309019dcf7ece0e921e6b926f)
- remove svelte config prerender.handleHttpError: warn (site now builds error free) [`f84171b`](https://github.com/janosh/matbench-discovery/commit/f84171bba2e88cee4ba9988839bd3cbfe0e06cec)
- add mdsvex custom img component [`2f6fae3`](https://github.com/janosh/matbench-discovery/commit/2f6fae3d6561c3c6201d6fc528f2d6d8aa4dc93f)
- add site/src/routes/paper/iclr-ml4mat/+page.md [`32ed170`](https://github.com/janosh/matbench-discovery/commit/32ed170d210727abafe3c25a9f8d93c0639adfa5)
- add /models page to site [`f28cc6d`](https://github.com/janosh/matbench-discovery/commit/f28cc6d283e1817660ce586fc0151bbe24e097e4)
- add kwarg metrics=("Precision", "Recall") to cumulative_precision_recall() [`4cf5028`](https://github.com/janosh/matbench-discovery/commit/4cf50282d4cc0e7dffcf08c5a5339eca3f8cac74)
- support plotly backend in rolling_mae_vs_hull_dist() [`fbf6a02`](https://github.com/janosh/matbench-discovery/commit/fbf6a02b723b04669070dd06fb67e6a2c7e78985)
- refactor plot func hist_classified_stable_as_func_of_hull_dist() [`967b482`](https://github.com/janosh/matbench-discovery/commit/967b482a10088ebfa946ea8df9a361ede76786b1)
- initial commit [`a111222`](https://github.com/janosh/matbench-discovery/commit/a111222c141474319482f78935cc6da716d678b3)
- add new module matbench_discovery/data.py with func load_wbm() [`0773112`](https://github.com/janosh/matbench-discovery/commit/077311216c62aecc3005da3719d7a16bbde8f555)
- add models/cgcnn/{slurm_train_cgcnn_ensemble,use_cgcnn_ensemble}.py [`a2e3f46`](https://github.com/janosh/matbench-discovery/commit/a2e3f46c3d8b4f31923b4a47529d5aa212dd507d)
- breaking add e_form_per_atm cutoff of +/- 5 eV/atom to WBM test set in fetch_process_wbm_dataset.py [`827c543`](https://github.com/janosh/matbench-discovery/commit/827c543f21a3e92270cbc59edd5eb257d7dd26aa)
- extract plot func rolling_mae_vs_hull_dist() from rolling MAE plot scripts [`8d65a7a`](https://github.com/janosh/matbench-discovery/commit/8d65a7aad620e2718012c03048b5b835a3c8a8db)
- fix WBM index mismatch between CSEs and summary [`20080b5`](https://github.com/janosh/matbench-discovery/commit/20080b5d9621f05ad558e24b92d923b994bdf3e1)
- refactor rolling_mae_vs_hull_dist() to support plotting multiple lines at once [`1b7f056`](https://github.com/janosh/matbench-discovery/commit/1b7f0560dca12724137208d1707530850c4ce02e)
- upgrade /models page with sorting and reporting model metrics (MAE, R2, Recall, ...) and stats (run time, date added, ...) [`8079ca1`](https://github.com/janosh/matbench-discovery/commit/8079ca1645de3a372b5ea2fa18208cd857178533)
- add backend=plotly | matplotlib to hist_classified_stable_vs_hull_dist() [`0fb7550`](https://github.com/janosh/matbench-discovery/commit/0fb755053e9dc36fe27d86fc2be89c33ec47b8f1)
- add route /about-the-test-set/tmi to display per-batch and per-arity WBM element counts [`5d98946`](https://github.com/janosh/matbench-discovery/commit/5d98946eb30a83be1b061d6abdaad74779240d6e)
- add classify_stable() in matbench_discovery/energy.py [`d593ae2`](https://github.com/janosh/matbench-discovery/commit/d593ae2586b059267b041ee29a762c4ecfbb0e13)
- add wandb.log scatter-parity plot in test_{cgcnn,wrenformer}.py [`255429a`](https://github.com/janosh/matbench-discovery/commit/255429a4c75369de633a2cdc41b3fc537e59df32)
- add slurm_submit_python() in new module mb_discovery/slurm.py [`97c6949`](https://github.com/janosh/matbench-discovery/commit/97c6949fcf9a3ec084bf6d4d7ae92f3ac1800825)
- use stable_metrics() from matbench_discovery.energy in scripts/compile_metrics.py and update site/src/routes/models/2023-01-23-model-stats.json [`75fc095`](https://github.com/janosh/matbench-discovery/commit/75fc0951779c10adcdb65fadf6474a8a0f5f6bdc)
- add MP (train set) elem count heatmap for comparison with WBM (test set) heatmap [`3130c89`](https://github.com/janosh/matbench-discovery/commit/3130c89ce408d6d16c3bad8183ec2fb55bce1489)
- crop WBM data to formation energy cutoff of +/-5 eV/atom (drops 524 materials, 502 below -5, 22 above +5) [`ad8349b`](https://github.com/janosh/matbench-discovery/commit/ad8349b17b9b322ad999ca38d4125bbcc4757edb)
- add ml_stability/m3gnet/slurm_array_m3gnet_relax_wbm.py [`1e96458`](https://github.com/janosh/matbench-discovery/commit/1e9645840f2241d5744a2f5290085f6344c37f5f)
- rename plot func precision_recall_vs_calc_count() to cumulative_clf_metric() that plots single metric at a time [`86f85f3`](https://github.com/janosh/matbench-discovery/commit/86f85f36488ec901b92b574f8307f0361bfa6295)
- fix scripts/hist_classified_stable_vs_hull_dist_models.py [`4ae422d`](https://github.com/janosh/matbench-discovery/commit/4ae422d71c548d1c13e8827a59cf3328d8c698b1)
- add models/bowsr/slurm_array_bowsr_megnet_relax_wbm.py [`e6bf955`](https://github.com/janosh/matbench-discovery/commit/e6bf9557daa0ae77116352225ec3fbccaa8cc10b)
- fix scripts/prc_roc_curves_models.py and roc-models.svelte fig [`a42472c`](https://github.com/janosh/matbench-discovery/commit/a42472c35073aa7b1a7b47e65a32c2bd9c66c218)
- add 2023-01-26-model-run-times-pie.svelte to /models page (from scripts/compile_metrics.py) [`0b8426a`](https://github.com/janosh/matbench-discovery/commit/0b8426abddf50100498050b9386a2b5f52828e5c)
- add test_precision_recall_vs_calc_count() [`e54a8c6`](https://github.com/janosh/matbench-discovery/commit/e54a8c6dd8b16f018c2b25e3b7f12b8652317cde)
- add mb_discovery/build_phase_diagram.py [`42a7909`](https://github.com/janosh/matbench-discovery/commit/42a7909bd753adacd67b5bf8d84691ab743f0104)
- fix legend/subplot titles in scripts/scatter_e_above_hull_models.py [`2f795f7`](https://github.com/janosh/matbench-discovery/commit/2f795f7c755865b326ed5cec203b291931249d7b)
- add precision_recall_vs_calc_count() to plot_funcs.py extracted from plot_scripts/precision_recall_vs_calc_count.py [`7ed1b09`](https://github.com/janosh/matbench-discovery/commit/7ed1b09da288dafdc403d1a201d9a8c360e05d5f)
- provisionally last set of fixes/refactors to data/wbm/fetch_process_wbm_dataset.py [`4936079`](https://github.com/janosh/matbench-discovery/commit/493607918bb4b49a6ca611b756e97822a4250660)
- add 2023-01-18-e-form-scatter-models.png to paper [`d1751a3`](https://github.com/janosh/matbench-discovery/commit/d1751a35c0e045f50cd7545d2ecea885d847cbd6)
- remove kwarg stability_crit from hist_classified_stable_vs_hull_dist() and cumulative_clf_metric() [`c5d3496`](https://github.com/janosh/matbench-discovery/commit/c5d34963b035d94d3f81d55ee087838751411bbc)
- add plot_scripts/all_models_scatter.py [`0236c36`](https://github.com/janosh/matbench-discovery/commit/0236c367171651c483b7a963d7342eca06ebfe21)
- add scripts/difficult_structures.py [`db34c09`](https://github.com/janosh/matbench-discovery/commit/db34c092502e557bb017b58e09116e25fd3299ec)
- make 2nd arg elemental_ref_entries to get_form_energy_per_atom() optional [`2517855`](https://github.com/janosh/matbench-discovery/commit/2517855ef971cabeecd0e796eaf27dd58c4d79a0)
- fix load_train_test() caching all data versions to same directory [`5084886`](https://github.com/janosh/matbench-discovery/commit/50848865f8a47dfdd45cad1c8d8b81fed0778f20)
- add citation.cff, add step-by-step guide for adding new models to the leaderboard to site/src/routes/how-to-contribute/+page.md [`4121f49`](https://github.com/janosh/matbench-discovery/commit/4121f4985008775d665b0f60843f27dccdf7e9b0)
- fix voronoi RF metadata.yml [`387722a`](https://github.com/janosh/matbench-discovery/commit/387722a006074c592354e8f1cfe2ad115baae891)
- add slurm submit and wandb logging to train_voronoi_rf.py [`6e58a1b`](https://github.com/janosh/matbench-discovery/commit/6e58a1b36390d4ca7b37b6ef152b76d9b9522bb3)
- eda_wbm_pre_vs_post_m3gnet_relaxation.py add code for 2 new plots [`bf36e42`](https://github.com/janosh/matbench-discovery/commit/bf36e423b9c67ff997a760fa5987abc170e74b39)
- SSoT for projects URLs in site/package.json [`a4b37a3`](https://github.com/janosh/matbench-discovery/commit/a4b37a352fe40f597db4c99ed5b561794321179d)
- add scripts/roc_models.py [`32a02d8`](https://github.com/janosh/matbench-discovery/commit/32a02d8232f888a90efd716b79d6a6a578bdb935)
- add rolling_accuracy kwarg and doc str to hist_classified_stable_as_func_of_hull_dist() [`9c6d61c`](https://github.com/janosh/matbench-discovery/commit/9c6d61c0de88cfc1799639d44fccad21b0b47747)
- add scripts/metrics_table.py [`20ef518`](https://github.com/janosh/matbench-discovery/commit/20ef518e6e7f4b79f01c62fb37efa4cdbc19f863)
- add data/wbm/compare_cse_vs_ce_mp_2020_corrections.py [`c1e55e1`](https://github.com/janosh/matbench-discovery/commit/c1e55e1ef3fad178b9ea949a69451152136032dc)
- add code to generate hist_classify_stable_as_func_of_hull_dist at end of ml_stability/m3gnet/m3gnet_relax_wbm.py [`383b8f4`](https://github.com/janosh/matbench-discovery/commit/383b8f46ec84419dd769066776fb2090605903b1)
- add kwarg show_dft_acc=False to rolling_mae_vs_hull_dist() [`9aebb08`](https://github.com/janosh/matbench-discovery/commit/9aebb08207ffade01a26723bee7163b909838533)
- split model pred loading from CSV into new module matbench_discovery/preds.py [`0e9e5dc`](https://github.com/janosh/matbench-discovery/commit/0e9e5dc6af1e156ad2f0b8a639f58af385eab82a)
- add structure perturbation to train_cgcnn.py [`6dd4398`](https://github.com/janosh/matbench-discovery/commit/6dd43982ac3ca8ae720ccd9d89fc19414f7a793e)
- move mb_discovery/{energy/__init__.py -> energy.py} and add tests/test_energy.py [`aec9933`](https://github.com/janosh/matbench-discovery/commit/aec99338305a54ee121acedd529eee5b971ca6a7)
- make model switching easy in hist_classified_stable_as_func_of_hull_dist_batches.py [`c487910`](https://github.com/janosh/matbench-discovery/commit/c487910787939f887ca24503305bade8b9c1201c)
- convert old data and plot file paths from rhys to new repo [`e61c8c7`](https://github.com/janosh/matbench-discovery/commit/e61c8c73682b7217b706b9d22a430a9897154a47)
- add write_html() and write test for it [`7a642d7`](https://github.com/janosh/matbench-discovery/commit/7a642d7fb818a6139c5495a87d759dcaf84c3128)
- return slurm environment variables from slurm_submit_python() [`444fb7f`](https://github.com/janosh/matbench-discovery/commit/444fb7fe9edbee75024f4b71acb5b41a8bc7440d)
- add data/2022-09-19-wbm-formation-energy+wyckoff.csv [`50b5f28`](https://github.com/janosh/matbench-discovery/commit/50b5f2833f46598d18561fadb6c54fcbb4136964)
- add cumulative F1 score to cumulative precision recall plot [`6e36278`](https://github.com/janosh/matbench-discovery/commit/6e3627801ab1cbf05de80c547bf0aba69736a70e)
- compare IS2RE and RS2RE energies in eda_wbm_pre_vs_post_m3gnet_relaxation.py [`90e1b89`](https://github.com/janosh/matbench-discovery/commit/90e1b89a7ddfda6f3d6527afb181b3f23e187655)
- refactor models/wrenformer/{mp,m3gnet_train_set}/slurm_submit.py to use slurm_submit_python() [`a19ca5d`](https://github.com/janosh/matbench-discovery/commit/a19ca5d8fd9c56eae28b38aedeba21e59d8e8c70)
- don't hydrate pseudo-relaxed structures in join_{bowsr,m3gnet}_results.py [`2073300`](https://github.com/janosh/matbench-discovery/commit/2073300eefce23055d53a443633c9b5ea2729f90)
- append '-debug' to job_name if debugging automatically [`f9c348c`](https://github.com/janosh/matbench-discovery/commit/f9c348ce8b1f3e2879590ef0a76a429828b76177)
- add models/megnet/slurm_array_megnet_wbm.py [`ffa075e`](https://github.com/janosh/matbench-discovery/commit/ffa075e648f08a6da943ca5a673ce40448749d15)
- add models/bowsr/join_bowsr_results.py [`e7f4582`](https://github.com/janosh/matbench-discovery/commit/e7f45826e59233dc50aa232bcd72e23e52225bbb)
- add mb_discovery/energy/slurm_e_above_hull.py [`4f6d853`](https://github.com/janosh/matbench-discovery/commit/4f6d853d4bc3004fbd943e60e0d5c5374cec62f9)
- more refactors in precision_recall_vs_calc_count.py and moving_hull_dist_mae_compare_models.py [`8daf748`](https://github.com/janosh/matbench-discovery/commit/8daf74855a97c330ef4da35b1f737de442dcefb3)
- add ml_stability/pre-vs-post-m3gnet-relaxation.py [`68a370a`](https://github.com/janosh/matbench-discovery/commit/68a370adf57f97faafe3db4ba0a0db25a4b68d51)
- add site Footer, add next/prev page links (with wrap-around) [`98a4289`](https://github.com/janosh/matbench-discovery/commit/98a4289e0830352132f14b35af83622a08736175)
- rename job_id->slurm_job_id in wandb run.config, rename 2022-08-16-m3gnet-wbm-relax-results-IS2RE.json.gz to 2022-08-16-m3gnet-wbm-IS2RE.json.gz [`0473994`](https://github.com/janosh/matbench-discovery/commit/047399422d703c18188d7ba0279eba2e5a3f2cea)
- update all figures with M3GNet + MEGNet and CGCNN=P results [`f9d4d04`](https://github.com/janosh/matbench-discovery/commit/f9d4d04726e718390a87a50b4b0433db45caf597)
- migrate setup.py to pyproject.toml [`81dc5f0`](https://github.com/janosh/matbench-discovery/commit/81dc5f036719d6ee0ee3507a6459583fd0ce65cc)
- rename project from ml-stability to matbench-discovery [`7c30b09`](https://github.com/janosh/matbench-discovery/commit/7c30b094d54bcc31c4e832bf5ef3de49e8a7c404)
- make PNG of WBM e_form hist in data/wbm/readme.md interactive plotly figure [`9223c75`](https://github.com/janosh/matbench-discovery/commit/9223c7556b9e87183c91d098cf18ab083ce8005c)
- fix misnamed variable in rolling_mae_vs_hull_dist() e_above_hull_pred->e_above_hull_error [`f5057ac`](https://github.com/janosh/matbench-discovery/commit/f5057ac0e1bb7e1252d44a90f27bb35cc34728f8)
- compare_cse_vs_ce_mp_2020_corrections.py code for <https://github.com/materialsproject/pymatgen/issues/2730> [`6101995`](https://github.com/janosh/matbench-discovery/commit/6101995c0eb8efd96ff97b2f195c238a95cd3b48)
- add mb_discovery/wrenformer/{mp,m3gnet_train_set} [`3d42214`](https://github.com/janosh/matbench-discovery/commit/3d42214b11a4644c60210e5dd128632958f2b017)
- add wandb scatter plot logging to slurm_array_megnet_wbm.py [`fed968f`](https://github.com/janosh/matbench-discovery/commit/fed968ff2768623574618946b387ad85d4341d13)
- add m3gnet to precision_recall_vs_calc_count.py script and save output pdf [`e776882`](https://github.com/janosh/matbench-discovery/commit/e776882faf33f9db271a03e0984673290b7b924f)
- add code to run test_cgcnn.py with task_type=RS2RE [`a2a99f2`](https://github.com/janosh/matbench-discovery/commit/a2a99f2323dd51e80b882563dbba3bf64466be88)
- fix recursion error when saving voronoi-tesselation to JSON [`38a9d7a`](https://github.com/janosh/matbench-discovery/commit/38a9d7afcef8e608146dcbdb453a2e59b21bae22)
- rename load_wbm() -> load_train_test() and add ability to download MP training files too [`811f581`](https://github.com/janosh/matbench-discovery/commit/811f581e5bfb9743d8245f51d9cb4c12c46bc83d)
- add tests/test_models.py [`ae020a7`](https://github.com/janosh/matbench-discovery/commit/ae020a7240dd6931416256a204d892c15bb8ed83)
- rename enrichment factor to DAF (discovery acceleration factor) [`1ee98fe`](https://github.com/janosh/matbench-discovery/commit/1ee98fe11326431d3cecea981b965f10ff6c4f4d)
- pnpm add -D katex remark-math for equation support in docs [`18be9dc`](https://github.com/janosh/matbench-discovery/commit/18be9dcb95ca3f79e143f6894717a78c382ea501)
- merge all /api/* routes into single page /api [`2908fdf`](https://github.com/janosh/matbench-discovery/commit/2908fdf5255e4cf1f7c02d5ddd07041750f2312b)
- add wandb_log_scatter() to plots.py, use in test_{wrenformer,cgcnn,megnet} [`65172ff`](https://github.com/janosh/matbench-discovery/commit/65172ffcc977bc4a099cd7dae43f4647c7ba638c)
- move model run/eval scripts from mb_discovery/ to new dir models/ [`89976c4`](https://github.com/janosh/matbench-discovery/commit/89976c4d10405999fd217c2ad7855273d2708c1c)
- add usage example code fence and WBM summary column glossary to how-to-use/+page.svx [`5405005`](https://github.com/janosh/matbench-discovery/commit/540500586e7cf753f780fb52a951d29f25234d50)
- add kwarg intersect_lines: str | Sequence[str] = () to precision_recall_vs_calc_count() [`39fa65a`](https://github.com/janosh/matbench-discovery/commit/39fa65a7422fa3554546f33dadeeddd340a9e2ba)
- refactor hist_clf_vary.py [`d5422bd`](https://github.com/janosh/matbench-discovery/commit/d5422bd5d9e4b7d7f6a5c90a58ded1e30ae9371c)
- add models/voronoi/join_voronoi_features.py [`b3c3aba`](https://github.com/janosh/matbench-discovery/commit/b3c3abaf56a12e96fe61bf4c6a8742732deb9fef)
- add models/voronoi/train_voronoi_rf.py [`65d1b54`](https://github.com/janosh/matbench-discovery/commit/65d1b54fd7c9b23b2049bc14bd2f545b8281e5aa)
- file tables and improved wording in data/wbm/readme.md [`d491439`](https://github.com/janosh/matbench-discovery/commit/d491439654a7ab3b2e7a710256546d9c19b20267)
- mv {mb->matbench}_discovery [`9fdab8c`](https://github.com/janosh/matbench-discovery/commit/9fdab8cb304b8d451104a21463ff86605172744d)
- update all plot scripts to use new 2022-10-19-wbm-summary.csv [`8d9e346`](https://github.com/janosh/matbench-discovery/commit/8d9e346d6bc52a509059cffa9fbe19dfe911b5fe)
- improve test_slurm_submit(), add tests/test_init.py [`da2337a`](https://github.com/janosh/matbench-discovery/commit/da2337a80f41eb0d4baf41943c1bab8b53ad892d)
- drop trajectories when joining slurm array m3gnet results [`c3c084b`](https://github.com/janosh/matbench-discovery/commit/c3c084b22677d2058a07360eba8dc6128ef4c5c1)
- add aflow wyckoff labels to 2022-10-19-wbm-summary.csv [`6450ebb`](https://github.com/janosh/matbench-discovery/commit/6450ebb45e3f9922f9628166cdc69bbcd0e0fc28)
- add scripts/hist_classified_stable_vs_hull_dist_models.py [`6030ef3`](https://github.com/janosh/matbench-discovery/commit/6030ef3b37d4aa8786b6bae1c121d42d84c8c6f5)
- add scripts/update_wandb_runs.py [`2779538`](https://github.com/janosh/matbench-discovery/commit/277953877f8e3eac8ee0c4f8ec9ae975de3610d6)
- add figures/2022-11-17-m3gnet-wbm-hull-dist-hist-batches.pdf [`de61b89`](https://github.com/janosh/matbench-discovery/commit/de61b89cde6ffc1e746aa1e88e0da55653ddf939)
- lower dft accuracy cone in hist_classified_stable_vs_hull_dist() [`8777870`](https://github.com/janosh/matbench-discovery/commit/877787046359e15bac05c234965b5d66be5e9f02)
- add test_rolling_mae_vs_hull_dist() to test_plot_funcs.py [`17df9d0`](https://github.com/janosh/matbench-discovery/commit/17df9d07eaada84567649918629ed105f6bab146)
- create no-structure CSVs of m3gnet and bowsr-megnet preds for faster loading used in load_df_wbm_with_preds() [`63b14aa`](https://github.com/janosh/matbench-discovery/commit/63b14aa82338366bd9344a2815ea798bac7e63f1)
- add slurm_submit_python() to use_cgcnn_ensemble.py and use_wrenformer_ensemble.py [`5d9a3b2`](https://github.com/janosh/matbench-discovery/commit/5d9a3b2b93e8df269ae6a4d37a9a1f28ed2cca3a)
- add models/voronoi/featurize_mp_wbm.py [`6932567`](https://github.com/janosh/matbench-discovery/commit/6932567ebb880f1084d1f6a9b5f8dfd21dd4180a)
- add slurm_submit_python() to featurize_mp_wbm.py [`784ee96`](https://github.com/janosh/matbench-discovery/commit/784ee96892c51149527d26339360418907d5f30e)
- make slurm_submit_python() kwargs partition, account non-optional [`d6c9a42`](https://github.com/janosh/matbench-discovery/commit/d6c9a4249f6b45ae019f040df0dd134bb8228a4f)
- record numpy + pytorch versions in train/test scripts [`7988f52`](https://github.com/janosh/matbench-discovery/commit/7988f52c580d220b0abb6b9c85df215e3973ac9a)
- slurm_submit() accept strings for slurm_flags, return timelimit, slurm_flags, pre_cmd as part of slurm_vars [`3a0173b`](https://github.com/janosh/matbench-discovery/commit/3a0173b1c5c2cb17e7f3956c2453f38fa77c7e3c)
- change plotly template to plotly_dark [`856c173`](https://github.com/janosh/matbench-discovery/commit/856c1736fc5ac754c7aeac448af3234e3cbd28e7)
- add GH actions that run pytest and markdown-link-check in CI [`92451da`](https://github.com/janosh/matbench-discovery/commit/92451da7ed00628d3d88616ab0405ec5d211f80a)
- rename e_above_mp_hull->e_above_hull_mp [`f25d91f`](https://github.com/janosh/matbench-discovery/commit/f25d91fd981f303b771dd8e1cee7d094f59b76af)
- centralize plot defaults into mb_discovery/plot_scripts/__init__.py [`f1cc667`](https://github.com/janosh/matbench-discovery/commit/f1cc6674c0847d7e37e1644527c6ee4c7fcda8d3)
- auto-regenerate API docs on production builds [`9a59191`](https://github.com/janosh/matbench-discovery/commit/9a5919149d64a4f301a5c950bf613ef2edc4c1a0)
- add df metadata to slurm_array_{m3gnet,bowsr}_wbm.py run_params [`f127a9e`](https://github.com/janosh/matbench-discovery/commit/f127a9ea60b971c54b61b0a50aca1959d5b0c917)
- set isort known_third_party = wandb [`da42037`](https://github.com/janosh/matbench-discovery/commit/da420373247db9beb5c0e3b96be61bdbd360a3db)
- wrap bowsr.optimize() in try/except [`8219c43`](https://github.com/janosh/matbench-discovery/commit/8219c4321dbe7811fc03d74981146b29720df0ce)
- more refactoring of rhys plotting scripts for clearer variable names [`a16b46a`](https://github.com/janosh/matbench-discovery/commit/a16b46a56b7a5083d4aa12ce12741cf87c252edf)
- fix models/bowsr/test_bowsr.py passing unrelaxed, not bowsr-relaxed structure to megnet [`0a2284a`](https://github.com/janosh/matbench-discovery/commit/0a2284ad14db848c500165c0203b56519a06809d)
- add m3gnet and bowsr-megnet to plot moving_hull_dist_mae_compare_models.py [`0ea1ff3`](https://github.com/janosh/matbench-discovery/commit/0ea1ff3d182e08e4babc95e3062f7d9fd9ea9109)
- add m3gnet/plots/2022-08-16-m3gnet-wbm-hull-dist-hist.pdf [`f50e1b3`](https://github.com/janosh/matbench-discovery/commit/f50e1b3dd9cf98be094f7fa3269c7e548756d1e9)
- fix models/voronoi/featurize_mp_wbm.py featurizer.set_n_jobs(1) [`8508c38`](https://github.com/janosh/matbench-discovery/commit/8508c38dafc07e22774d445fe4f9270a23896799)
- fix landing page clf metrics plot not applying inline styles [`89a33b3`](https://github.com/janosh/matbench-discovery/commit/89a33b3eaf58d8a8b9c1e3b47e8eeff0cef3a101)
- update wrenformer preds CSV path [`612d308`](https://github.com/janosh/matbench-discovery/commit/612d308e776d0d74f0790cb0d98bb0f6e7137a4d)
- fix slurm_array_m3gnet_relax_wbm.py wandb project m3gnet->matbench-discovery [`ba932c2`](https://github.com/janosh/matbench-discovery/commit/ba932c2038730586d68f09a3441a1b950edf86f5)
- matbench_discovery/plot_scripts/{moving_hull_dist_mae_compare_models -> rollng_mae_vs_hull_dist_all_models}.py [`b292e9b`](https://github.com/janosh/matbench-discovery/commit/b292e9b334803b61382adfa57288d1a15e9bfbcd)
- rename data files -all-mp-.../-mp-all-... to just -mp- [`9a41f7b`](https://github.com/janosh/matbench-discovery/commit/9a41f7ba86571935ad3dfd9a07874d507d7e09e1)
- add task_type variable to toggle between relaxing initial and final WBM structures in slurm_array_m3gnet_relax_wbm.py [`7006822`](https://github.com/janosh/matbench-discovery/commit/7006822ed5dcb30fdee1eae4b5c38f6d0219edf9)
- fix predict_from_wandb_checkpoints() didn't return ensemble_metrics due to missing target_col [`5804d13`](https://github.com/janosh/matbench-discovery/commit/5804d13b4b6dd2e4f8cc0a38aee691f9cdca3e43)
- mkdir data/mp and move some PPD pickles and MP CSEs in there [`8a8aac6`](https://github.com/janosh/matbench-discovery/commit/8a8aac65886dfcc2fce89a3b08a35364b89516c9)
- fix bad import in train_cgcnn [`1859e85`](https://github.com/janosh/matbench-discovery/commit/1859e858be8bb6aa2da92a0d906efec161c22669)
- matbench_discovery/__init__.py define CHECKPOINT_DIR = f"{ROOT}/wandb/checkpoints" [`4f1e5b6`](https://github.com/janosh/matbench-discovery/commit/4f1e5b6a8ead955af00730213f8271997100b5be)
- add ml_stability/m3gnet/plots/2022-08-02-m3gnet-wbm-hull-dist-hist.pdf [`1c3cc92`](https://github.com/janosh/matbench-discovery/commit/1c3cc92bf1f7c36c157bb1e1d9d71a0ef2cf2c63)
- slurm_submit_python() print sbatch command at submit time and in slurm job [`7b0918a`](https://github.com/janosh/matbench-discovery/commit/7b0918acd76b19cbeafac61ddadb64430399f0a2)
- remove wandb.login() calls, seems enough to do it once when initially running on new machine [`bac8551`](https://github.com/janosh/matbench-discovery/commit/bac85510f815d1a698e3f253ede9ec9660b95c7e)
- replace figshare wbm-steps-summary.csv imports with data/wbm/2022-10-19-wbm-summary.csv [`73655f6`](https://github.com/janosh/matbench-discovery/commit/73655f6051f4f7ac84f7a69d9fbc025528d8dbc5)
- mv matbench_discovery/plot_scripts -> scripts [`5860a5e`](https://github.com/janosh/matbench-discovery/commit/5860a5e52bec8b117ef708f1d28d18076227a9ab)
- migrate matbench_discovery/build_phase_diagram.py to new data/wbm/2022-10-19-wbm-cses+init-structs.json.bz2 [`f5e755e`](https://github.com/janosh/matbench-discovery/commit/f5e755eb85cc1ce843e421e9e207f029a6879967)
- fix path issue on case-sensitive file systems [`5e9f5df`](https://github.com/janosh/matbench-discovery/commit/5e9f5df44b4e238948e41760db9efc9d1791d954)
- rename col e_above_{->mp_}hull in data/2022-06-11-from-rhys/wbm-e-above-mp-hull.csv [`105e468`](https://github.com/janosh/matbench-discovery/commit/105e468f05afe24c1f8b7cd3049ef3ec1e175fd3)
- fill readme with project goals [`c369d18`](https://github.com/janosh/matbench-discovery/commit/c369d1819f789afbde2674e9ca11bfc4e50f7e49)
- improve package structure [`2be7bc3`](https://github.com/janosh/matbench-discovery/commit/2be7bc399955691b93867e9980ed7a3fc80ba12e)
- more code clean up [`b957855`](https://github.com/janosh/matbench-discovery/commit/b95785543f4c8a5cb24f8f1ba931b040d22724d8)
- rename hist_classified_stable_as_func_of_hull_dist() to hist_classified_stable_vs_hull_dist() [`24cab3d`](https://github.com/janosh/matbench-discovery/commit/24cab3d236fa8a36cc3e473c63d4ee4f970b9830)
- set python_requires>=3.9 so we can use | operator for dict merge [`03c6694`](https://github.com/janosh/matbench-discovery/commit/03c6694243a923645e77071494a6710a8e1457de)
- git rm --cached -r data/2022-06-11-from-rhys/ [`44d5133`](https://github.com/janosh/matbench-discovery/commit/44d5133769b1a572cb331a6048286c4d7804fc60)
- add test dep pytest-markdown-docs to test python code fences in markdown/svx docs [`359dfae`](https://github.com/janosh/matbench-discovery/commit/359dfae52a224f3df12dab9b1529ae32a9107edd)
- slurm_submit_python() don't print cmd if running slurm interactively [`3b2be74`](https://github.com/janosh/matbench-discovery/commit/3b2be7417a9d57fe2807e89fded77249362db8bf)
- mv ml_stability/{stability_plot_scripts->plots} [`c2a45b3`](https://github.com/janosh/matbench-discovery/commit/c2a45b3968a2904e5d3be7668ef76a71b8b8f66d)
- update file paths to CSV files in data/2022-06-11-from-rhys/ [`a9645b7`](https://github.com/janosh/matbench-discovery/commit/a9645b7aadbca26b29d8af0b94a5c3969f62e577)
- remove figures/ dir [`bf7f53a`](https://github.com/janosh/matbench-discovery/commit/bf7f53af2fe47e168b203663995faec332d51f2b)
- tweak plt.rc params [`1fb2149`](https://github.com/janosh/matbench-discovery/commit/1fb214922a7f6d6ef92892d07cf63e095db66fa4)
- dissolve dir models/wrenformer/m3gnet_train_set [`8acba18`](https://github.com/janosh/matbench-discovery/commit/8acba18161ab16487a3506fd8d75f4a11ed61aee)
- big file rename: all model scripts now named train|test_{model_name}.py [`06911dc`](https://github.com/janosh/matbench-discovery/commit/06911dcf151ded05a82205d24a8723158cc0fa90)
- add several new figures [`88dd6b4`](https://github.com/janosh/matbench-discovery/commit/88dd6b426322f669f8fed35ce0c2a54485572d44)
- data/2022-06-11-from-rhys/process_wbm_cleaned.py add warning about baked-in energy corrections in rhys' ComputedStructureEntries [`e5b099c`](https://github.com/janosh/matbench-discovery/commit/e5b099c54e632bffa08d7564bcc518fbbffb786f)
- improve readability of moving MAE scripts [`e13d624`](https://github.com/janosh/matbench-discovery/commit/e13d6244ebf83e5309480b493252eb7c72af118f)