1 Release published by 1 person

• v4.6.0

  published Feb 15, 2025

13 Pull requests merged by 2 people

• [docs] add a note about DART in early_stopping docstring

  #6846 merged Mar 4, 2025

• [ci] [dask] test lightgbm.dask on macOS

  #6677 merged Mar 2, 2025

• [ci] consolidate R package installs into a CI script

  #6808 merged Feb 24, 2025

• [ci] [R-package] add CI job using clang-20

  #6837 merged Feb 20, 2025

• [R-package] make finalize() in R6 classes private

  #6833 merged Feb 17, 2025

• bump development version to v4.6.0.99

  #6826 merged Feb 16, 2025

• release v4.6.0

  #6796 merged Feb 14, 2025

• [R-package] require lgb.Dataset, remove support for passing 'colnames' and 'categorical_feature' for lgb.train() and lgb.cv()

  #6714 merged Feb 13, 2025

• [python-package] support sub-classing scikit-learn estimators

  #6783 merged Feb 12, 2025

• [c++] update to fmt 11.1.2, fast_double_parser 0.8.0

  #6802 merged Feb 10, 2025

• [ci] fix valgrind workflow

  #6816 merged Feb 10, 2025

• [ci] validate pyproject file by json schema

  #6813 merged Feb 9, 2025

• [ci] [python-package] update pre-commit hooks to latest versions

  #6817 merged Feb 7, 2025

6 Pull requests opened by 5 people

• Add support to specifying position using position_column parameter

  #6825 opened Feb 15, 2025

• [c++] enhance error handling for forced splits file loading

  #6832 opened Feb 16, 2025

• [ci] build aarch64 wheel builds on a real aarch64 machine

  #6843 opened Feb 28, 2025

• [R-package] Allow for valid data to be a slice of the training data

  #6844 opened Feb 28, 2025

• [R-package] Ensure print.lgb.Booster() works when objective is not explicitly set

  #6850 opened Feb 28, 2025

• [python-package] Load parameters from model string

  #6852 opened Mar 2, 2025

13 Issues closed by 5 people

• [python-package] Early Stopping does not work in LGBMClassifier

  #6244 closed Mar 4, 2025

• [CLI] segfaults when training

  #6845 closed Feb 28, 2025

• [dask] Dask estimators not supported on Windows

  #6842 closed Feb 26, 2025

• Custom MAE Python Loss Function Does Not Match C++ Results

  #6744 closed Feb 21, 2025

• [R-package] {lightgbm} fails on r-hub clang20 container

  #6836 closed Feb 20, 2025

• [python-package] Unexpected behaviour if `pyarrow` is installed, but `cffi` is not

  #6782 closed Feb 18, 2025

• [R-package] CRAN requires a new release by Feb 14th

  #6791 closed Feb 18, 2025

• Can i load model from JSON format text which dump by using LGBM_BoosterDumpModel

  #6834 closed Feb 18, 2025

• [python-package] [CUDA] installing CUDA version with 'pip' hangs, consume all the memory (32 G) and swap space

  #6824 closed Feb 15, 2025

• [RFC] remove 'categorical_feature' and 'feature_name' parameters in cv() and train()

  #6435 closed Feb 13, 2025

• Disabling threading completely for prediction (in C++)

  #6820 closed Feb 10, 2025

• [python-package] How to copy a booster in Python?

  #6821 closed Feb 10, 2025

• Exception thrown in lib_lightgbm.dll on version v4.5.0

  #6779 closed Feb 9, 2025

20 Issues opened by 16 people

• [R-package] [ci] gcc14 test jobs failing: "libRlapack.so: undefined symbol: dgemmtr_"

  #6855 opened Mar 7, 2025

• [python-package] Load a lightgbm.LGBMRegressor from file

  #6854 opened Mar 7, 2025

• Shrinkage different on first tree between default and custom objective

  #6853 opened Mar 4, 2025

• [python-package] Model parameters are lost when loading Booster from model serialized string

  #6851 opened Mar 2, 2025

• [python-package] Fitting on Polars Dataframe fails due to missing setter for fetures_names_in_

  #6849 opened Feb 28, 2025

• [R-package] Printing a booster trained with lgb.train() errors unexpectedly

  #6848 opened Feb 28, 2025

• [CUDA] Potential missed inversion of discounts within CUDA implementation of LambdarankNDCG

  #6847 opened Feb 28, 2025

• Training crashes when `"monotone_constraints_method"` is set to `"advanced"`

  #6841 opened Feb 26, 2025

• [CUDA] Defaulting to malloc in CHAllocator

  #6840 opened Feb 24, 2025

• arrow.h brings lots of C++ to the c_api.h header

  #6839 opened Feb 20, 2025

• What does `booster.refit` actually do?

  #6838 opened Feb 20, 2025

• CUDA backend uses significantly more memory than OpenCL

  #6835 opened Feb 19, 2025

• [python-package] lgb.train() init_model does not allow a model that has fewer features than the training data

  #6831 opened Feb 16, 2025

• [c++] forcedsplits_filename pointing at a non-existent file is silently ignored

  #6830 opened Feb 15, 2025

• [ci] [python-package] [R-package] support MUSL-based Linux distributions (like Alpine)

  #6829 opened Feb 15, 2025

• [RFC] [CUDA] distribute Python wheels with CUDA support

  #6828 opened Feb 15, 2025

• NuGet package is missing README

  #6827 opened Feb 15, 2025

• How is NDCG computed for queries where there is only one document?

  #6823 opened Feb 11, 2025

• [python-package] [dask] AttributeError: 'Future' object has no attribute 'get_params()' in DaskLGBMClassifier.fit when using Dask Gateway

  #6822 opened Feb 11, 2025

• What happens to missing data during training when `use_missing=False`?

  #6819 opened Feb 7, 2025

27 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• [ROCm] add support for ROCm/HIP device

  #6086 commented on Mar 4, 2025 • 14 new comments

• [docs] update Python-package installation guide

  #6767 commented on Mar 2, 2025 • 2 new comments

• [cpp] fix: to_if_else missing check for threshold when missing_type is not none

  #6818 commented on Feb 8, 2025 • 0 new comments

• [ci] update libboost version from `1.74` to `1.83`

  #6582 commented on Feb 12, 2025 • 0 new comments

• [ci] Add cpp tests compiled via gcc on Linux

  #6315 commented on Feb 17, 2025 • 0 new comments

• [c++] Initial Work for Pairwise Ranking

  #6182 commented on Feb 18, 2025 • 0 new comments

• [CUDA] Multi-GPU for CUDA Version

  #6138 commented on Mar 7, 2025 • 0 new comments

• WIP: [python-package] preserve params when copying Booster (fixes #5539)

  #6101 commented on Feb 10, 2025 • 0 new comments

• [R-package] cannot uses slices of the same Dataset as training and validation datasets

  #6008 commented on Mar 7, 2025 • 0 new comments

• [ci] [python-package] [dask] dask tests failing: unexpected tuple return value somewhere

  #6739 commented on Mar 2, 2025 • 0 new comments

• [CUDA] illegal memory access when using CUDA and large max_bin and large dataset

  #6512 commented on Feb 28, 2025 • 0 new comments

• [ci] use real aarch64 runners instead of emulation with QEMU

  #6788 commented on Feb 28, 2025 • 0 new comments

• Rank Evaluation Metric NDCG inconsistent between LGBMRanker.fit vs. LGBMRanker.predict()

  #6814 commented on Feb 27, 2025 • 0 new comments

• [python-package] How do I reproduce LightGBM's L2 loss with a custom objective?

  #6040 commented on Feb 26, 2025 • 0 new comments

• Feature fraction gives different column selections when using a custom objective

  #6053 commented on Feb 26, 2025 • 0 new comments

• [CUDA] "Warning: 'Using sparse features with CUDA is currently not supported' in LightGBM with GPU"

  #6725 commented on Feb 25, 2025 • 0 new comments

• [dask] Dask LightGBM best practices

  #5840 commented on Feb 24, 2025 • 0 new comments

• light gbm hangs when loading a model file in subprocess

  #6137 commented on Feb 23, 2025 • 0 new comments

• [python-package] merge multiple data files into one

  #6151 commented on Feb 20, 2025 • 0 new comments

• [python-package] Failed to train model with dataset built incrementally

  #6770 commented on Feb 20, 2025 • 0 new comments

• [RFC] provide Python/R implementations of all the built-in objectives?

  #6440 commented on Feb 19, 2025 • 0 new comments

• Feature Requests & Voting Hub

  #2302 commented on Feb 15, 2025 • 0 new comments

• Feature Request: add timeout parameter to the .fit() method

  #6596 commented on Feb 15, 2025 • 0 new comments

• [GPU] [R-package] Installing a GPU-enabled Build in R --- amd gpu.

  #6786 commented on Feb 12, 2025 • 0 new comments

• [python-package] `Booster.refit()` segfaults when label only contains 1 class

  #6792 commented on Feb 10, 2025 • 0 new comments

• [GPU] GOSS boosting error on GPU H100

  #6811 commented on Feb 9, 2025 • 0 new comments

• [python] Reusing dataset constructed with free_raw_data = True isn't possible. Intended behaviour?

  #4965 commented on Feb 9, 2025 • 0 new comments