monkey0head comments

Repositories
Issues
Comments

Results 15 comments of


                                            monkey0head

create CandidatesGenerator

fix indexing: _fit and _predict with external ids filtering (inside scenario) were used by first level models to avoid redundant dataset reindexing. need to consider using _fit_wrap and _predict_wrap without...

notebook with ann inference example added

added ids conversion with StringIndexer instead of protected ._convert_index()

check for shuffle

Leaving max_seen items before join was an optimisation which I did. It speeds up shuffle significantly, because without filtering all data (num users in predict * num items in dataset)...

check for shuffle

[pref_analysis_predict_ml20_als.ipynb.txt](https://github.com/sberbank-ai-lab/RePlay/files/7703250/pref_analysis_predict_ml20_als.ipynb.txt)

Neural Click Models

Thank you for your contribution! Please, have a look at the comments and add a time measurements to the notebook to show the speed of the main stages of simulation...