Domain: AI Economics & Strategy

110 terms

Saddle Point Intermediate

A point where gradient is zero but is neither a max nor min; common in deep nets.
Scaling Laws Intermediate

Empirical laws linking model size, data, compute to performance.
Second-Order Methods Intermediate

Optimization using curvature information; often expensive at scale.
Self-Reflection Intermediate

Models evaluating and improving their own outputs.
Sharp Minimum Intermediate

A narrow minimum often associated with poorer generalization.
Sparse Attention Intermediate

Attention mechanisms that reduce quadratic complexity.
State Space Intermediate

All possible configurations an agent may encounter.
Stress Testing Intermediate

Simulating adverse scenarios.
Supply Chain Attack Intermediate

Compromising AI systems via libraries, models, or datasets.
Throughput Ceiling Intermediate

Maximum system processing rate.
Token Budgeting Intermediate

Limiting inference usage.
Toolformer Intermediate

Models trained to decide when to call tools.
Training Cost Intermediate

Cost of model training.
Universal Approximation Theorem Intermediate

Neural networks can approximate any continuous function under certain conditions.
Use-Case Classification Intermediate

Categorizing AI applications by impact and regulatory risk.
Value at Risk Intermediate

Maximum expected loss under normal conditions.
Value Function Intermediate

Expected cumulative reward from a state or state-action pair.
Variance Term Intermediate

Error due to sensitivity to fluctuations in the training dataset.
VC Dimension Intermediate

A measure of a model class’s expressive capacity based on its ability to shatter datasets.
Warmup Intermediate

Gradually increasing learning rate at training start to avoid divergence.