What is Dueling DQN?
Machine Learning / AI
What architectural trends are present in modern decoder-only LLMs (LLaMA, Mistral, Qwen)?
How else can an agent be evaluated besides LLM-as-a-Judge: what criteria can be used to understand if it works correctly or close to correct?
What neural network libraries have you used?
What are the bottlenecks when deploying GNN in production?
How does BERT capture the semantic similarity between sentences that are written differently but have the same meaning?
Which option is better for transmitting large text data (about 10 GB)?
What is your higher education degree and field of study?
We are solving a linear regression problem, all y > 0. Which of the listed algorithms can give a negative prediction: linear regression, KNN, gradient boosting, tree, random forest, neural network?
What is One-Class SVM?
What is an agent's long-term memory and how to build it (summarization, vector memory)?
What is MAE?
How is the threshold selected and what criterion is used?
What is Bayesian Optimization and acquisition functions (EI, UCB, PI)?
What is TensorRT-LLM and where does it have advantages?
What is a hash collision and what methods exist to resolve it?
Why is it not enough to just calculate the median on each worker and then combine?
Is there a way to verify that generated text does not contain factual errors relative to the source?
What is Bayesian target encoding?
How do you assess your learning ability and digital thinking?