benfred / implicit

Fast Python Collaborative Filtering for Implicit Feedback Datasets
https://benfred.github.io/implicit/
MIT License
3.57k stars 612 forks source link

Early stopping for ALS #714

Open characat0 opened 6 months ago

characat0 commented 6 months ago

Following on https://github.com/benfred/implicit/issues/9, we could add the ability to stop on certain conditions defined by the user (every k steps, and on metrics that are quick to compute), right now in my company we hacked a solution for early stopping by using the callback function passed when fitting a model. We can work on a PR with a more clean approach for this feature, any thoughts on this?