search

sberbank-ai-lab / RePlay

RecSys Library
https://sberbank-ai-lab.github.io/RePlay/
Apache License 2.0
68 stars 6 forks source link

fallback scenario indexers issue #69

Closed monkey0head closed 2 years ago

monkey0head commented 2 years ago

In addition to #68 there is probably a bug in fallback scenarios indexers. they seem to be one object, but they need to be different objects for proper cold users and items processing. length of scenario user indexer labels is not equal to the number of users in train dataset. also new users from test were not included to the indexers (at least poprec indexers inside Fallback), but should be. I believe, the length of user indexer for fb_model in scenario should be 6040 as for pure poprec model. image

Darel13712 commented 2 years ago

They are different indexers. One thing to remember is that main and fallback models do not train on full dataset, but only on a warm part of it. Only cold_model sees the whole dataset. https://github.com/sberbank-ai-lab/RePlay/blob/ed6598aaaec03238c60a86d4c213e0f584272da5/replay/scenarios/basescenario.py#L39

monkey0head commented 2 years ago

closed by #96