Vocabulary
Understand the term. Connect the concept.
A practical dictionary of 512 AI, machine learning, mathematics, and programming terms from HerbDev’s AI Vocabulary app. Start with a definition, explore an example, then follow the concepts behind it.
Linked concepts open in a new tab. Audio reads the definition and example.
Start with the concepts that come first.
Each path places prerequisites before the concepts that depend on them. Open a term, explore its example, and use its recall question before moving on.
AI and machine learning foundations · 11 concepts
- Artificial Intelligence
Computing systems designed to perform tasks associated with human intelligence.
- Dataset
A collection of examples organized for analysis or learning.
- Feature
An input attribute or derived representation used by a model.
- Model
A mathematical or computational representation used to explain or predict behavior.
- Machine Learning
Methods that fit patterns from data to improve performance on a task.
- Label
A provided target associated with a training or evaluation example.
- Supervised Learning
Learning a mapping from inputs to provided target outputs.
- Training Set
Examples used to fit model parameters.
- Test Set
Examples reserved for final evaluation after development choices are fixed.
- Generalization
How well a learned model works on relevant unseen data.
- Overfitting
Learning training-specific patterns that fail to transfer well.
How models learn · 13 concepts
- Scalar
A single numerical value rather than an array of values.
- Weight
A learned coefficient controlling an input's contribution to a computation.
- Bias
A learned offset added to a computation independently of its current input.
- Function
A rule or callable that maps inputs to outputs or performs a computation.
- Activation Function
A transformation applied to a unit's signal, often to introduce nonlinearity.
- Neuron
A computational unit that combines inputs and applies a transformation.
- Neural Network
A parameterized model built from connected computational units or layers.
- Derivative
The local rate at which a function's output changes with its input.
- Partial Derivative
A derivative with respect to one variable while holding other inputs fixed.
- Vector
An ordered collection of numbers representing a point or direction.
- Gradient
The vector of a scalar function's partial derivatives.
- Learning Rate
The scale applied to a gradient-based parameter update.
- Gradient Descent
An iterative method that moves parameters opposite the gradient to reduce an objective locally.
Evaluate a model · 15 concepts
- Dataset
A collection of examples organized for analysis or learning.
- Feature
An input attribute or derived representation used by a model.
- Model
A mathematical or computational representation used to explain or predict behavior.
- Machine Learning
Methods that fit patterns from data to improve performance on a task.
- Label
A provided target associated with a training or evaluation example.
- Supervised Learning
Learning a mapping from inputs to provided target outputs.
- Classification
Predicting a discrete class or class probabilities.
- Accuracy
The fraction of evaluated predictions that are correct.
- True Positive
A positive prediction whose actual label is positive.
- False Positive
A positive prediction whose actual label is negative.
- Precision
The fraction of predicted positives that are truly positive.
- False Negative
A negative prediction whose actual label is positive.
- Recall
The fraction of actual positives that the system identifies.
- Probability
A numerical measure of uncertainty assigned to an event.
- Calibration
Agreement between predicted probabilities and observed outcome frequencies.
Understand language models · 21 concepts
- Token
A discrete unit in a model's input or output sequence.
- Scalar
A single numerical value rather than an array of values.
- Vector
An ordered collection of numbers representing a point or direction.
- Embedding
A numerical representation placing items in a vector space.
- Query
An attention representation used to score candidate keys.
- Key
An attention representation compared with queries to determine relevance scores.
- Value
An attention representation combined using attention weights.
- Probability
A numerical measure of uncertainty assigned to an event.
- Softmax
A transformation from real-valued scores to a positive vector summing to one.
- Attention
A mechanism that weights information from candidate sources based on relevance.
- Self-Attention
Attention in which queries, keys, and values come from the same sequence representation.
- Positional Encoding
Information added or applied to represent positions or order.
- Transformer
A neural architecture built around attention and feature transformations.
- Language Model
A model of probabilities or scores over language sequences.
- Model
A mathematical or computational representation used to explain or predict behavior.
- Dataset
A collection of examples organized for analysis or learning.
- Function
A rule or callable that maps inputs to outputs or performs a computation.
- Loss Function
A rule assigning a numerical penalty to predictions or model behavior.
- Training
Adjusting a model using data and a learning objective.
- Pretraining
Initial training that builds reusable model capabilities before later adaptation.
- Large Language Model
A language model trained at substantial scale in data and model capacity.
Representations and agent memory · 6 concepts
- Scalar
A single numerical value rather than an array of values.
- Vector
An ordered collection of numbers representing a point or direction.
- Embedding
A numerical representation placing items in a vector space.
- Contextual Embedding
A representation whose value depends on the surrounding input.
- Dataset
A collection of examples organized for analysis or learning.
- Memory
Stored information made available to later agent steps or sessions.
Manage browser learning data
“Remove my test data” clears all AI 101 and ML 101 scores, reviews, project checklists and rating selections, plus saved vocabulary and recent history in this browser. Test entries cannot be separated from other learning progress. Cookie preferences, security protections, submitted feedback, contact emails and past analytics are kept.