Difficulty: Intermediate

412 terms

Spurious Correlation Intermediate

Model relies on irrelevant signals.

Stability Intermediate

System returns to equilibrium after disturbance.

State Space Intermediate

All possible configurations an agent may encounter.

State Space Model Intermediate

Models time evolution via hidden states.

Stochastic Approximation Intermediate

Optimization under uncertainty.

Stochastic Gradient Descent Intermediate

A gradient method using random minibatches for efficient training on large datasets.

Stress Testing Intermediate

Simulating adverse scenarios.

Structured Output Intermediate

Forcing predictable formats for downstream systems; reduces parsing errors and supports validation/guardrails.

Supervised Learning Intermediate

Learning a function from input-output pairs (labeled data), optimizing performance on predicting outputs for unseen inputs.

Supply Chain Attack Intermediate

Compromising AI systems via libraries, models, or datasets.

Synthetic Data Intermediate

Artificially created data used to train/test models; helpful for privacy and coverage, risky if unrealistic.

System Prompt Intermediate

A high-priority instruction layer setting overarching behavior constraints for a chat model.

Temperature Intermediate

Scales logits before sampling; higher increases randomness/diversity, lower increases determinism.

Temporal Convolution Intermediate

CNNs applied to time series.

Text-to-Speech Intermediate

Generating speech audio from text, with control over prosody, speaker identity, and style.

Throughput Intermediate

How many requests or tokens can be processed per unit time; affects scalability and cost.

Throughput Ceiling Intermediate

Maximum system processing rate.

Time Series Intermediate

Sequential data indexed by time.

Token Budgeting Intermediate

Limiting inference usage.

Tokenization Intermediate

Converting text into discrete units (tokens) for modeling; subword tokenizers balance vocabulary size and coverage.

Tool Use Intermediate

Letting an LLM call external functions/APIs to fetch data, compute, or take actions, improving reliability.

Toolformer Intermediate

Models trained to decide when to call tools.

Top-k Intermediate

Samples from the k highest-probability tokens to limit unlikely outputs.

Top-p Intermediate

Samples from the smallest set of tokens whose probabilities sum to p, adapting set size by context.

Train/Validation/Test Split Intermediate

Separating data into training (fit), validation (tune), and test (final estimate) to avoid leakage and optimism bias.

Training Cost Intermediate

Cost of model training.

Training Pipeline Intermediate

End-to-end process for model training.

Transfer Learning Intermediate

Reusing knowledge from a source task/domain to improve learning on a target task/domain, typically via pretrained models.

Transformer Intermediate

Architecture based on self-attention and feedforward layers; foundation of modern LLMs and many multimodal models.

Transparency Obligation Intermediate

Requirement to inform users about AI use.