v0.8.1

What's Changed

Neural Networks

• Remove torchtext and update dependencies (#22)

Contributors

Thanks all for the contributions :)
@cjlin1 @Eleven1Liu

v0.8.0

What's Changed

Linear tree-based models

• Update build_tree function with SparseKmeans implementation (#19)

Bugs Fixing 🐛 and Code Enhancement ✨

• Upgrade minimum python version from 3.8 to 3.10 and remove deprecated APIs (i.e., numpy, spicy) (#20)
• Fix the bug in pruning tree predictions (#23)

Contributors

Thanks all for the contributions :)
@cjlin1 @khoinpd0411 @maclin726 @zhi-bao @Eleven1Liu

New Contributors

Welcome @khoinpd0411 ➿

v0.7.4

What's Changed

Highlights

• Update inference method involving prune tree prediction ( #11)

Bugs Fixing 🐛

• Add psutil dependency for PyPi installation (#15)

Contributors

Thanks all for the contributions :)
@cjlin1 @zhi-bao

v0.7.3

What's Changed

Linear tree-based models

• Avoid additional memory usage when performing hstack (#3)
• Apply weight recalculation when using L1/L2-SVM dual solver (#10)

Bugs Fixing 🐛 and Code Enhancement ✨

• Fix the primal solver issue (#12)
• Unify doc strings (#13)

Contributors

Thanks all for the contributions :)
@cjlin1 @zhi-bao @maclin726 @will945945945 @Eleven1Liu

v0.7.2

⭐️⭐️ Restart LibMultiLabel in @ntumlgroup 🏡

What's Changed

Documents

• Update the repository link on the LibMultiLabel webpage and make minor UI adjustments (#4, #6)

Linear tree-based models

• Estimate tree model size (#385)
• Change the default optimizer to l2 cddual in the example configurations for l2svm: EUR-Lex/tree_l2svm.yml, rcv1/l2svm.yml, rcv1/l2svm_svm_format.yml (#388)
• Do not filter any data instance in the root (#1)
• Keep model outputs as probability estimates (#2)

Fix APIs / Bugs 🐛

• Fix bugs in latest version of NumPy (#7, #8)
• [linear] Explicitly assign arguments to avoid incorrect argument assignments (#389)

Contributors

Thanks all for the contributions :)
@cjlin1 @ericliu8168 @poppopting @ChengYehLi @maclin726 @will945945945 @Eleven1Liu

v0.7.1

What's Changed

Documents

• Update docs by revising --no_merge_train_val to --merge_train_val (#368)
• Add multi-label tutorial (#381)

Fix APIs / Bugs 🐛

• Fix GridSearchCV to support sklearn>=1.3.0 (#362)
• Update regex syntax for python3.12 (#373)
• Update Ray Argument (#376)
• Fix Autotest: Set Pytorch <=2.3 to resolve torchtext dependency issue (#386)

Contributors

Thank you all for your contributions :)
@cjlin1 @donglihe-hub @Gordon119 @thomas0125 @poppopting @Eleven1Liu

New Contributors

Welcome @thomas0125 ✨ @poppopting 🏜️🐪

Original release date: Aug 22, 2024

v0.7.0

What's Changed

Documents

• Add Handling Data with Many Labels with AttentionXML (#368)

Highlights

• Add AttentionXML (#346)
• Add new argument beam_width (#366, #367)
• Remove unused elements in nn.model (#363, #369)

Fix APIs 🔧

• Use the latest protocol in pickle dump (#364)

Contributors

Thank you all for your contributions :)
@cjlin1 @donglihe-hub @Sinacam @will945945945 @Gordon119 @Eleven1Liu

Original release date: May 24, 2024

v0.6.2

What's Changed

Fix APIs 🔧

• Fixed sklearn API with version >=1.3.0 (#362)

Contributors

Thank you all for your contributions :)
@cjlin1 @thomas0125 @Sinacam @will945945945

New Contributors

Welcome @thomas0125 ✨

Original release date: Mar 7, 2024

v0.6.1

What's Changed

Fix APIs 🔧

• Fixed deprecated/removed scipy API (#360)
• Fixed sklearn version before GridSearchCV is fixed (#361)

Contributors

Thank you all for your contributions :)
@cjlin1 @Sinacam

Original release date: Feb 28, 2024

v0.6.0

What's Changed

Highlights

• Update package versions: torch>=2.0.1 and lightning==2.0.9 with CUDA 11.8 and 12.1
• For users using torch 1.13.1 with CUDA 11.6 or 11.7, please git checkout 10819d0 or pip install PyPi==0.5.2.

Contributors

Thank you all for your contributions :)
@cjlin1 @Eleven1Liu @Sinacam @Gordon119 and all lab members who provided helpful feedback!

Original release date: Feb 10, 2024