Difficulty: Intermediate
Model relies on irrelevant signals.
System returns to equilibrium after disturbance.
All possible configurations an agent may encounter.
Models time evolution via hidden states.
Optimization under uncertainty.
A gradient method using random minibatches for efficient training on large datasets.
Simulating adverse scenarios.
Forcing predictable formats for downstream systems; reduces parsing errors and supports validation/guardrails.
Learning a function from input-output pairs (labeled data), optimizing performance on predicting outputs for unseen inputs.
Compromising AI systems via libraries, models, or datasets.
Artificially created data used to train/test models; helpful for privacy and coverage, risky if unrealistic.
A high-priority instruction layer setting overarching behavior constraints for a chat model.
Scales logits before sampling; higher increases randomness/diversity, lower increases determinism.
CNNs applied to time series.
Generating speech audio from text, with control over prosody, speaker identity, and style.
How many requests or tokens can be processed per unit time; affects scalability and cost.
Maximum system processing rate.
Sequential data indexed by time.
Limiting inference usage.
Converting text into discrete units (tokens) for modeling; subword tokenizers balance vocabulary size and coverage.
Letting an LLM call external functions/APIs to fetch data, compute, or take actions, improving reliability.
Models trained to decide when to call tools.
Samples from the k highest-probability tokens to limit unlikely outputs.
Samples from the smallest set of tokens whose probabilities sum to p, adapting set size by context.
Separating data into training (fit), validation (tune), and test (final estimate) to avoid leakage and optimism bias.
Cost of model training.
End-to-end process for model training.
Reusing knowledge from a source task/domain to improve learning on a target task/domain, typically via pretrained models.
Architecture based on self-attention and feedforward layers; foundation of modern LLMs and many multimodal models.
Requirement to inform users about AI use.