Releases: automl/Auto-PyTorch
Releases · automl/Auto-PyTorch
v0.2.1
- [FIX] ADD forecasting init design to pip data files by @dengdifan in #459
- checks for time series dataset split by @dengdifan in #464
- [FIX] Numerical stability scaling for timeseries forecasting tasks by @dengdifan in #467
- [FIX] pipeline options in
fit_pipeline
by @ravinkohli in #466 - [FIX] results management and visualisation with missing test data by @ravinkohli in #465
- [ADD] Robustly refit models in final ensemble in parallel by @ravinkohli in #471
Full Changelog: v0.2...0.2.1
Version 0.2
What's Changed
- Fix 361 by @ravinkohli in #367
- [ADD] Test evaluator by @ravinkohli in #368
- [fix] Hotfix debug no training in simple intensifier by @nabenabe0928 in #370
- [fix] Change int to np.int32 for the ndarray dtype specification by @nabenabe0928 in #371
- [ADD] variance thresholding by @ravinkohli in #373
- [ADD] scalers from autosklearn by @ravinkohli in #372
- [FIX] Remove redundant categorical imputation by @ravinkohli in #375
- [feat] Add coalescer by @nabenabe0928 in #376
- Fix: keyword arguments to submit by @eddiebergman in #384
- [FIX] Datamanager in memory by @ravinkohli in #382
- [feat] Add new task inference for APT by @nabenabe0928 in #386
- [fix] Update the SMAC version by @nabenabe0928 in #388
- [ADD] dataset compression by @ravinkohli in #387
- [refactor] Fix SparseMatrixType --> spmatrix and add ispandas by @nabenabe0928 in #397
- [ADD] feature preprocessors from autosklearn by @ravinkohli in #378
- [feat] Add str to autoPyTorchEnum by @nabenabe0928 in #405
- [ADD] Subsampling Dataset by @ravinkohli in #398
- fix dist twine check for github by @dengdifan in #439
- Time series forecasting by @dengdifan in #434
- fit updates in gluonts by @dengdifan in #445
- docs for forecasting task by @dengdifan in #443
- [ADD] Allow users to pass feat types to tabular validator by @ravinkohli in #441
- [RELEASE] Changes for release v0.2 by @ravinkohli in #446
- [FIX] Documentation and docker workflow file by @ravinkohli in #449
- [ADD] change log for release by @ravinkohli in #450
- [RELEASE] Release v0.2 by @ravinkohli in #448
New Contributors
- @dengdifan made their first contribution in #439
Full Changelog: v0.1.1...v0.2
v0.1.1
What's Changed
- fixed README by @urbanmatthias in #1
- Update develop branch to 0.0.2 release by @LMZimmer in #17
- set fill value to max of full dataset + 1 by @jonathanburns in #36
- Include missing files in sdist by @thatch in #47
- sort value range sets before adding as hyperparameter by @ntnguyen-dev in #43
- Image classification full cs by @dwoiwode in #73
- Document formatting by @daikikatsuragawa in #64
- Allow specifying the network type in include by @franchuterivera in #78
- Search space update by @ravinkohli in #80
- Network Cleanup by @bastiscode in #81
- Make sure the performance of pipeline is at least 0.8 by @franchuterivera in #82
- Refactor development docs by @franchuterivera in #83
- Feature preprocessors, Loss strategies by @ravinkohli in #86
- Handling Input to auto pytorch by @franchuterivera in #89
- Adding tabular regression pipeline by @bastiscode in #85
- FIX weighted loss issue by @ravinkohli in #94
- Reduce Deadlock Probability by @franchuterivera in #84
- handle nans in categorical columns by @ravinkohli in #118
- Pre fetch openml data for pytest by @franchuterivera in #112
- Embedding layer by @ravinkohli in #91
- Fixes to address automlbenchmark problems by @franchuterivera in #126
- Bug fix in Test API by @ravinkohli in #129
- Refactoring base dataset by @nabenabe0928 in #105
- move to a minimization problem by @franchuterivera in #113
- FIX_123 by @franchuterivera in #133
- Adds more examples to customise AutoPyTorch. by @ravinkohli in #124
- [Feat] Better traditional pipeline cutoff time by @franchuterivera in #141
- Hyperparameter Search Space updates now with constant and include ability by @ravinkohli in #146
- [Bug] Fix random halt problems on traditional pipelines by @franchuterivera in #147
- Run history traditional by @ravinkohli in #121
- [FIX] Enables backend to track the num run by @franchuterivera in #162
- [Doc] First push of the developer documentation by @franchuterivera in #127
- Refactoring base dataset splitting functions by @nabenabe0928 in #106
- [Fix] Refactor development reproducibility by @franchuterivera in #172
- [ADD] Extra visualization example by @franchuterivera in #189
- [Fix] docs links by @franchuterivera in #201
- [Refactor] Use the backend implementation from automl common by @franchuterivera in #185
- [DOC] Adds documentation to the abstract evaluator by @franchuterivera in #160
- [FIX] Update Readme by @franchuterivera in #208
- Reduce run time of the test by @ravinkohli in #205
- [refactor] Getting dataset properties from the dataset object by @ravinkohli in #164
- Change ubuntu version in docs workflow by @ravinkohli in #237
- Add dist check worflow by @ravinkohli in #238
- [feature] Greedy Portfolio by @ravinkohli in #200
- [ADD] Forkserver as default multiprocessing strategy by @franchuterivera in #223
- [ADD] Get incumbent config by @ravinkohli in #175
- [ADD] Coverage calculation by @franchuterivera in #224
- [ADD] Pytest schedule by @ravinkohli in #234
- [fix] Dropout bug fix by @ravinkohli in #247
- [FIX] Fixes for Tabular Regression by @ravinkohli in #235
- [doc] Add the bug fix information for ipython user by @nabenabe0928 in #254
- [ADD] Enable long running regression by @franchuterivera in #251
- [MAINT] Drop 3.6 python support by @franchuterivera in #258
- [ADD] Stricter checks mypy by @ravinkohli in #240
- [feat] Add flexible step-wise LR scheduler with minimum changes by @nabenabe0928 in #256
- Enable python 3.9 by @franchuterivera in #264
- Fixes problems with weighted cross-entropy by @franchuterivera in #263
- [Fix] long running regression by @franchuterivera in #272
- [Fix] budget allocation to enable runtime/epoch as budget by @franchuterivera in #271
- [fix] Be able to display error messages in additional info as it is by @nabenabe0928 in #225
- [ADD] Add column transformer by @ravinkohli in #305
- [FIX] Minor Fixes by @ravinkohli in #306
- [Add] Dockerfile by @ravinkohli in #314
- [style] Remove prefix
typing
and adapt to google style doc by @nabenabe0928 in #307 - [ADD] Missing Batchnorm by @ravinkohli in #317
- [feat] Update automl_common and add setup.py for submodule by @nabenabe0928 in #324
- [FIX] Additional metrics during training by @ravinkohli in #316
- [ADD] Update developer documentation by @ravinkohli in #320
- [FIX] remove .pth for early stopping by @ravinkohli in #321
- [Add] documentation and example for parallel computation by @ravinkohli in #322
- [ADD] Documentation for data validation and preprocessing by @ravinkohli in #323
- [ADD] documentation for pipelines and steps by @ravinkohli in #329
- Final changes for v0.1.0 by @ravinkohli in #341
New Contributors
- @urbanmatthias made their first contribution in #1
- @jonathanburns made their first contribution in #36
- @thatch made their first contribution in #47
- @ntnguyen-dev made their first contribution in #43
- @dwoiwode made their first contribution in #73
- @daikikatsuragawa made their first contribution in #64
- @bastiscode made their first contribution in #81
Full Changelog: v0.0.2...v0.1.1
v0.0.2: Merge pull request #15 from automl/release_0.0.2
The first pre-alpha of Auto-PyTorch supporting image data