Difficulty: Intermediate

412 terms

Model Moat Intermediate

Competitive advantage from proprietary models/data.
Model Orchestration Intermediate

Coordinating models, tools, and logic.
Model Predictive Control Intermediate

Optimizes future actions using a model of dynamics.
Model Registry Intermediate

Central system to store model versions, metadata, approvals, and deployment state.
Model Release Control Intermediate

Restricting distribution of powerful models.
Model Risk Intermediate

Risk of incorrect financial models.
Model Risk Management Intermediate

Framework for identifying, measuring, and mitigating model risks.
Model Stealing Intermediate

Reconstructing a model or its capabilities via API queries or leaked artifacts.
Model Tiering Intermediate

Classifying models by impact level.
Model Watermarking Intermediate

Embedding signals to prove model ownership.
Momentum Intermediate

Uses an exponential moving average of gradients to speed convergence and reduce oscillation.
Monitoring Intermediate

Observing model inputs/outputs, latency, cost, and quality over time to catch regressions and drift.
Multi-Agent System Intermediate

Multiple agents interacting cooperatively or competitively.
Multi-Head Attention Intermediate

Allows model to attend to information from different subspaces simultaneously.
Multimodal Fusion Intermediate

Combining signals from multiple modalities.
Multimodal Model Intermediate

Models that process or generate multiple modalities, enabling vision-language tasks, speech, video understanding, etc.
Multitask Learning Intermediate

Training one model on multiple tasks simultaneously to improve generalization through shared structure.
Mutual Information Intermediate

Quantifies shared information between random variables.
Neural Network Intermediate

A parameterized function composed of interconnected units organized in layers with nonlinear activations.
Neural Vocoder Intermediate

Generates audio waveforms from spectrograms.
Next-Token Prediction Intermediate

Training objective where the model predicts the next token given previous tokens (causal modeling).
NIST AI RMF Intermediate

US framework for AI risk governance.
NLP Intermediate

AI subfield dealing with understanding and generating human language, including syntax, semantics, and pragmatics.
Non-Convex Optimization Intermediate

Optimization with multiple local minima/saddle points; typical in neural networks.
Normalization Intermediate

Techniques that stabilize and speed training by normalizing activations; LayerNorm is common in Transformers.
Object Detection Intermediate

Identifying and localizing objects in images, often with confidence scores and bounding rectangles.
Objective Function Intermediate

A scalar measure optimized during training, typically expected loss over data, sometimes with regularization terms.
Objective Surface Intermediate

Visualization of optimization landscape.
Observability Intermediate

A broader capability to infer internal system state from telemetry, crucial for AI services and agents.
Off-Policy Learning Intermediate

Learning from data generated by a different policy.