Machine Learning / AI

Have you worked directly with PostgreSQL?

What do you forget to do with data on CUDA that can lead to memory overflow during training?

What happens if you do select 1 instead of a column name?

What is supervised fine-tuning (SFT)? What data is needed?

Middle — Senior

ТензорВижнТехМагнит

What problems arise if batch normalization updates statistics during validation?

What is change point detection?

What is dropout and how is it implemented?

How does Tableau collect data: independently or through connection to a data warehouse?

What is DPO and why is it more popular than PPO in open models?

Middle — Senior

СамокатМагнит

How to approach object detection tasks with severe data imbalance?

How does temperature affect the probability distribution in softmax?

How to select backbones for classification?

Why is it important to look for similar layers in teacher and student during distillation?

Why is the number of values in the results of a query with aggregate functions not equal to the number of rows?

What is a residual connection and what is it used for?

Middle

ДомкликМагнит

What is tool-calling and what types of tools do agents use?

What three matrices are used in multihead attention?

Why is it necessary to set the model to inference mode (model.eval()) before testing?

What is the difference between key padding mask and attention mask in PyTorch?

What is BatchNorm?