CHANGELOG.md

Changelog

HEAD

Features

• Check learner parameters during the model construction.

1.2.0 - 2023-01-24

Features

• Add support for distributed training and distributed hyper-parameter tuning in the OSS build. See https://www.tensorflow.org/decision_forests/distributed_training
• Setting "subsample" is enough enable random subsampling (to need to also set "sampling_method=RANDOM").
• Add "min_vocab_frequency" argument in "FeatureUsage" to control the minimum frequency of categorical items.
• Add "override_global_imputation_value" argument in "FeatureUsage" to override the value used for global imputation of missing value by the global-imputation algorithm.
• The Tuner argument "use_predefined_hps" automatically configures the set of hyper-parameters to explore during automatic hyper-parameter tuning.
• Replaces the MEAN_MIN_DEPTH variable importance with INV_MEAN_MIN_DEPTH.
• Add option to forbid model inference with the slow inference engine.

Fix

• Automatic documentation generation for RandomForestModel and other classes.

1.1.0 - 2022-11-18

Features

• Support for Tensorflow Serving APIs.
• Add support for zipped Yggdrasil Decision Forests model for yggdrasil_model_to_keras_model.
• Added model prediction tutorial.
• Prevent premature stopping of GBT training through new parameter early_stopping_initial_iteration.

Fix

• Using loaded datasets with TF-DF no longer fails (Github #131).
• Automatically infer the semantic of int8 values as numerical (was categorical before).
• Build script fixed
• Model saving no longer fails when using invalid feature names.
• Added keyword to pandas dataset drop (Github #135).

1.0.1 - 2022-09-07

Fix

• Issue in the application of auditwheel in TF 1.0.0.

1.0.0 - 2022-09-07

Features

• Add customization of the number of IO threads when using fit_on_dataset_path.

Fix

• Improved documentation
• Improved testing and stability

0.2.7 - 2022-06-15

Features

• Multithreading of the oblique splitter for gradient boosted tree models.
• Support for pure serving model i.e. model containing only serving data.
• Add "edit_model" cli tool.

Fix

• Remove bias toward low outcome in uplift modeling.

0.2.6 - 2022-05-17

Features

• Support for TensorFlow 2.9.1

0.2.5 - 2022-05-17

Features

• Adds the contrib module for contributed, non-core functionality.
• Adds contrib.scikit_learn_model_converter, which facilitates converting Scikit-Learn tree-based models into TF-DF models.
• Discard hessian splits with score lower than the parents. This change has little effect on the model quality, but it can reduce its size.
• Add internal flag hessian_split_score_subtract_parent to subtract the parent score in the computation of an hessian split score.
• Add support for hyper-parameter optimizers (also called tuner).
• Add text pretty print of trees with tree.pretty() or str(tree).
• Add support for loading YDF models with file prefixes. Newly created models have a random prefix attached to them. This allows combining multiple models in Keras.
• Add support for discretized numerical features.

0.2.4 - 2021-02-04

Features

• Support for TensorFlow 2.8.

0.2.3 - 2021-01-27

Features

• Honest Random Forests (also work with Gradient Boosted Tree and CART).
• Can train Random Forests with example sampling without replacement.
• Add support for Focal Loss with Gradient Boosted Trees.
• Add support for MacOS.

Fixes

• Incorrect default evaluation of categorical split with uplift tasks. This was making uplift models with missing categorical values perform worst, and made the inference of uplift model possibly slower.
• Fix pd_dataframe_to_tf_dataset on Pandas dataframe not containing arrays.

0.2.2 - 2021-12-13

Features

• Surface the validation_interval_in_trees, keep_non_leaf_label_distribution and 'random_seed' hyper-parameters.
• Add the batch_size argument in the pd_dataframe_to_tf_dataset utility.
• Automatically determine the number of threads if num_threads=None.
• Add constructor argument try_resume_training to facilitate resuming training.
• Check that the training dataset is well configured for TF-DF e.g. no repeat operation, has a large enough batch size, etc. The check can be disabled with check_dataset=False.
• When a model is created manually with the model builder, and if the dataspec is not provided, tries to adapt the dataspec so that the model looks as if it was trained with the global imputation strategy for missing values (i.e. missing_value_policy: GLOBAL_IMPUTATION). This makes manually created models more likely to be compatible with the fast inference engines.
• TF-DF models fit method now passes the validation_data to the Yggdrasil learners. This is used for example for early stopping in the case of GBT model.
• Add the "loss" parameter of the GBT model directly in the model constructor.
• Control the amount of training logs displayed in the notebook (if using notebook) or in the console with the verbose constructor argument and fit parameter of the model.

Fixes

• num_candidate_attributes is not ignored anymore when num_candidate_attributes_ratio=-1.
• Use the median bucket split value strategy in the discretized numerical splitters (local and distributed).
• Surface the max_num_scanned_rows_to_accumulate_statistics parameter to control how many examples are scanned to determine the feature statistics when training from a file dataset with fit_on_dataset_path.

0.2.1 - 2021-11-05

Features

• Compatibility with TensorFlow 2.7.0.

0.2.0 - 2021-10-29

Features

• Add advanced option predict_single_probability_for_binary_classification to generate prediction tensors of shape [batch_size, 2] for binary classification model.
• Add support for weighted training.
• Add support for permutation variable importance in the GBT learner with the compute_permutation_variable_importance parameter.
• Support for tf.int8 and tf.int16 values.
• Support for distributed gradient boosted trees learning. Currently, the TF ParameterServerStrategy distribution strategy is only available in monolithic TF-DF builds. The Yggdrasil Decision Forest GRPC distribute strategy can be used instead.
• Support for training from dataset stored on disk in CSV and RecordIO format (instead of creating a tensorflow dataset). This option is currently more efficient for distributed training (until the ParameterServerStrategy support per-worker datasets).
• Add max_vocab_count argument to the model constructor. The existing max_vocab_count argument in FeatureUsage objects take precedence.

Fixes

• Missing filtering of unique values in the categorical-set training feature accumulator. Was responsible for a small (e.g. ~0.5% on SST2 dataset) drop of accuracy compared to the C++ API.
• Fix broken support for max_vocab_count in a FeatureUsage with type CATEGORICAL_SET.

0.1.9 - 2021-08-31

Features

• Disable tree pruning in the CART algorithm if the validation dataset is empty (i.e. validation_ratio=0).
• Migration to Tensorflow 2.6. You will see an undefined symbol error if you install this version with a TensorFlow version different than 2.6. Previous versions were compiled for TF 2.5.

Fixes

• Fix failure from Github Issue #45 where the wrong field was accessed for leaf node distributions.
• Fix saving of categorical features specification in the Builder.

0.1.8 - 2021-07-28

Features

• Model can be composed with the functional Keras API before being trained.
• Makes all the Yggdrasil structural variable importances available.
• Makes getting the variable importance instantaneous.
• Surface the name argument in the model classes constructors.
• Add a postprocessing model constructor argument to easy apply post-processing on the model predictions without relying on the Keras Functional API.
• Add extract_all_trees method in the model inspector to efficiently exact all the trees.
• Add num_threads constructor argument to control the number of training threads without using the advanced configuration.
• By default, remove the temporary directory used to train the model when the model python object is garbage collected.
• Add the import_dataspec constructor argument to the model builder to import the feature definition and dictionaries (instead of relying on automatic discovery).

Changes

• When saving a model in a directory already containing a model, only the assets directory is entirely removed before the export (instead of the entire model directory).

Fixes

• Wrong label shape in the model inspector's objective field for pre-integerized labels.

0.1.7 - 2021-06-23

Features

• Add more of characters to the non-recommended list of feature name characters.
• Make the inference op multi-thread compatible.
• Print an explicit error and some instructions when training a model with a Pandas dataframe.
• pd_dataframe_to_tf_dataset can automatically rename feature to make them compatible with SavedModel export signatures.
• model.save(...) can override an existing model.
• The link function of GBT model can be removed. For example, a binary classification GBT model trained with apply_link_function=False will output logits.

0.1.6 - 2021-06-07

Features

• Add hyper-parameter sorting_strategy to disable the computation of the pre-sorted index (slower to train, but consumes less memory).
• Format wrapper code for colab help display.
• Raises an error when a feature name is not compatible (e.g. contains a space).

0.1.5 - 2021-05-26

Features

• Raise an error of the number of classes is greater than 100 (can be disabled).
• Raise an error if the model's task does not match the pd_dataframe_to_tf_dataset's task.

Bug fix

• Fix failure when input feature contains commas.

0.1.4 - 2021-05-21

Features

• Stop the training when interrupting a colab cell / typing ctrl-c.
• model.fit support training callbacks and a validation dataset.

Bug fix

• Fix failure when there are not input features.

0.1.3 - 2021-05-19

Features

• Register new inference engines.

0.1.2 - 2021-05-18

Features

• Inference engines: QuickScorer Extended and Pred

0.1.1 - 2021-05-17

Features

• Migration to TensorFlow 2.5.0.
• By default, use a pre-compiled version of the OP wrappers.

Bug fix

• Add missing plotter.js from Pip package.
• Use GitHub version of Yggdrasil Decision Forests by default.

0.1.0 - 2021-05-11

Initial Release of TensorFlow Decision Forests.

Features

• Random Forest learner.
• Gradient Boosted Tree learner.
• CART learner.
• Model inspector: Inspect the internal model structure.
• Model plotter: Plot decision trees.
• Model builder: Create model "by hand".