What is CUTLASS?
Machine Learning / AI
How to split data into strata for validation?
How do mean/max pooling differ from learnable pooling?
What is vLLM and why is PagedAttention needed?
What is a cross-encoder and how does it differ from a bi-encoder?
How to organize human eval of LLM responses in a team?
What is a model registry and what tools have you used (MLflow, W&B)?
What is the role of residual connection and LayerNorm in a Transformer block?
How is the probability distribution of a normal distribution arranged in binary classification?
What is an advantage function and why does it reduce variance?
What types of data quality issues have you encountered?
Why is stratification metrics needed?
What is Dueling DQN?
What are DINO and DINO-v2?
What classification metrics should be used in case of severe class imbalance?
What is the logic behind a recommendation system?
What are A100, H100, B200, and their key architectural differences?
What is a safety classifier at the end of a dialogue?
What is IVF index and why are centroids needed?
What metrics and methods are used to evaluate models and AB testing?