How can hyperparameters be tuned?
Machine Learning / AI
What approximate salary range are you aiming for?
Tell us about TF-IDF: how it works, what the final matrix looks like.
What is a graph RAG and in what tasks is it effective?
Как учитывать временной промежуток между покупками для релевантности товаров?
Which metric should be chosen for regression with outliers: MSE or MAE?
What business metrics and counter-metrics should be measured to evaluate a product promotion widget?
How to use collaborative methods for recommendations based on embeddings?
What splitting criteria are used in decision trees (e.g., entropy criterion)? Can you name others?
What loss function is used by default in linear regression?
Tell us about regularization in classical ML algorithms: L1 and L2. What are their properties and differences?
Why is the analytical formula for linear regression not used in practice?
How does stochastic gradient descent differ from the classical one?
Why is it necessary to use the LEAD function to find future values by id in SQL?
How will we tune the parameters in gradient boosting? What parameters are there and what will you prioritize in tuning?
How does dropout work during training and inference?
What error threshold should be used to stop training?
How are coroutines and the event loop structured in Python?
What are your future wishes, what do you want to do?