Building Applications with AI Agents

Glossary

Activation function

A mathematical function applied to a neuron’s input to determine the output, such as ReLU (rectified linear unit), Sigmoid, or Softmax.

Agent

An autonomous system that can perform tasks, make decisions, and interact with users or environments.

Artificial intelligence (AI)

The simulation of human intelligence in machines, including reasoning, learning, and problem-solving.

Attention mechanism

A technique in neural networks that allows models to focus on specific parts of input sequences, crucial for transformers.

Backpropagation

The algorithm used to train neural networks by adjusting weights based on error gradients.

Beam search

A decoding algorithm in sequence generation models that selects the most probable sequences.

Bias

Systematic errors in machine learning models that can lead to unfair outcomes.

Chatbot

An AI system that interacts with users via natural language.

Cold start problem

The challenge of making predictions when little or no historical data is available.

Context window

The amount of text (measured in tokens) an LLM can process at once.

Corpus

A collection of text data used to train or fine-tune language models.

Decoder

The component in transformer-based models that generates output sequences from encoded information.

Dense vector

A numerical representation of text in vector space, often used in embeddings.

Domain adaptation

The process of fine-tuning a model to perform better on a specific domain.

Dropout ...