Recbole

* Highlights
* New Features
* Bug Fixes
* Code Refactor

Highlights

The RecBole v0.2.1 release includes a number of wonderful new features, some bug fixes and code refactor. In this version, we pay more attention to improving user experience. A few of the highlights include:

1. We add 7 new models into RecBole.
2. We add colors to logger and now RecBole is "colorful".
3. `Dataset` and `Dataloader` can be saved now, which makes RecBole much more flexible.
4. Now you can get training loss line graph of models by set `draw_loss_pic` .

New Features

* Add 7 new models:
* General recommendation(6) : EASE(609), RaCT(732), RecVAE(727), NNCF(713), ENMF(643), SLIMElastic(621)
* External algorithm lib model(1) : LightGBM (685)
* We add color to logger info, which makes logger much more clear (761)
* We add `plot_train_loss()` in trainer, and now user can get training loss line graph of model(724)
* We add `dataset.save()` and `save_split_dataloader()`, and now users can save pre_processed dataset or pre_processed dataloaders and reload them for other models training. (760)
* We add other parameters (including model parameters) output in logger (725)
* We add example code of case study and save/load in `run_example/` (774)
* We add `docs/` into RecBole (735)

Bug Fixes

* Fix a datatype bug in Windows, which may cause runtime error when run sequential models in Windows platform（710）

* Fix a bug in `general_dataloader`, which may cause runtime error when `ContextFullDataLoader` is empty (723)

Code Refactor

* Refactor the `eval_setting.py` (743)
* Refactor the `data_preparation` (751)

* Highlights
* New Features
* Improvements
* Bug Fixes

Highlights

The RecBole v0.2.0 release includes a number of new features, model efficiency improvements and bug fixes. A few of the highlights include:

1. We add 12 new models into RecBole, including several **non-sampling models** and an external algorithm lib model: **XGBoost**.
2. **Case study** is added to RecBole, which is helpful for users to analyze the model result (e.g: give an item ID and a user ID and get the score and ranking position of the item).
3. We improve the efficiency of data loading and negative sampling.
4. We now support the full ranking evaluation for context-aware recommendation models.

New Features

* Add 12 new models:
* Sequential recommendation(6) : RepeatNet(\584), Fossil(\580), HGN(575), SHAN(573), NPE(573)，HRM(573)
* General recommendation(5) : CDAE(642), MacridVAE(637), MultiVAE(626), MultiDAE(626), LINE(\591)
* External algorithm lib model(1) : XGBoost(\557)
* ``training_neg_sample_num`` of pairwise_loss model now can be greater than `1` (533).
* We add `training_neg_sample_distribution` in config setting to choose the negative sampling strategy during training (534).
* We add `benchmark_filename` in config setting to load pre-split dataset (\596).
* Progress bar is added for training and evaluating (\618).
* We add `loss_decimal_place` and `metric_decimal_place` in config setting to control decimal place of loss and metric results separately (625).
* We add `GAUC` metric into evaluation (reference: [Deep Interest Network for Click-Through Rate Prediction](https://arxiv.org/abs/1706.06978), KDD 2018) (#572).
* We add `unused_col` in config setting to drop the columns only used in data preparation but not used in model (559).

Improvements

* We support the ranking evaluation for context-aware recommendation models (503).
* We improve the efficiency of data loading and negative sampling (559).
* We remove the `pre_neg_sampling` in Dataloader, which is helpless to model training (559).
* We improve the underlying data structure of RecBole, which can promote efficiency of data processing (559).
* We refactor the evaluation code (572) and reformat the mode code (647).

Bug Fixes

Model

* Fix a bug in NeuMF model: this bug may cause `dropout_ratio` disable (629).
* Fix a bug in NGCF and GCMC model: now the sparse dropout is disable during evaluation in NGCF and GCMC (601).
* Fix a bug in DCN model: this bug may cause crash when running on CPU (633).
* Fix a bug in BERT4Rec model: this bug may cause crash when running on CPU (556).

Trainer

* Fix a bug in the `Trainer._generate_train_loss_output()`: this bug may cause the training log is missing (559).

Data

* Fix a bug in the sampler: this bug may cause runtime error (559)

* Highlights
* New Features
* Improvements
* Bug Fixes

Highlights

The RecBole v0.1.2 release includes a number of new features, model efficiency improvements and bug fixes. A few of the highlights include:

1. We add CI (Continuous Integration) to RecBole, which improves the efficiency and quality of our development.
2. We improve the efficiency of GNN-based general recommendation models(NGCF, GCMC, LightGCN, SpectralCF) by refactoring the construction of sparse interaction tensor. In this way, usage of GPU RAM can be greatly reduced.
3. Some bugs in models, trainer and data are fixed.

New Features

* Add gradient clipping. (533)
* Add `Dataset`'s attributes `token2id` and `id2token`.(511)
* Add continuous integration. (496)

Improvements

* Improve the efficiency of GNN-based general recommendation models (NGCF, GCMC, LightGCN, SpectralCF). (525, 526)
* `SequentialDataloader`: add support for historical sequence of Non-SEQ features. (547)

Bug Fixes

Model:

* Fix a bug in the DMF model: crash when training set contains item without interactive records. ( 505)

* Fix a bug in the TransRec model: this bug may cause runtime error. (502)

Trainer:

* Fix a bug in the trainer: this bug may cause the NeuMF model to report an error if `eval_batch_size=1`.(537)

* Fix the bug that some models don't work when `epochs = 0`. (499)

Data:

* Fix wrong data type of time field in `SequentialDataloader`. (551)

* Fix the bug of negative sample judgment in sampler. (551)

* Fix the bug that `kg_feat` is empty when `KGDataloader` is in KGRS mode. (551）

Basic RecBole