Daria

### Feature Description - Add optional argument `model_meta` to `MetricsApp.construct` method - `model_meta` is pd.DataFrame with obligatory `Columns.Model` columns. All other are model params - Join models meta and models...

### Feature Description `use_gpu` argument to `ImplicitRanker.rank` method that calls `implicit.gpu.KnnQuery().topk` instead of `implicit.cpu.topk.topk` when set to True. ### Why this feature? Allows for gpu inference for a range of...

SasRec sequential recommender model in Fit/Recommend paradigm.

ModelLike pipeline with candidates generation, targets creation, sampling negatives, feature collection and reranker training. Fully compatible with RecTools Fit/Recommend paradigm.

### What happened? Precision (and other Calssification and Ranking metrics) is not correct when there are duplicates in test interactions. See screenshot ### Expected behavior I suppose that correct value...

### Feature Description A property that will return full learnt embedding size for vector models. For ials it will count latent + feature factors. For lightfm it will add biases....

### Feature Description Attribute that specifies if model uses features when fitting. Another option is to allow user to choose if he wants model to use features. Yet to be...

### Feature Description `take_features` method will accept item features selection and user features selection an return a `Dataset` with only those features that are specified ### Why this feature? -...

### Feature Description Wrappers initialisation from full list of hyper-params (including wrapped) ### Why this feature? - It will help us for save/load functions to models. - It will help...