AI & ML Terms reference data
Key terminology in artificial intelligence and machine learning.
Topic guide
What AI & ML Terms covers
AI and machine-learning reference data covers the vocabulary around large language models and the methods under them: how models are structured (architectures, attention), how they are trained (optimizers, training algorithms, regularization, transfer learning), how they are evaluated (metrics) and the concepts that shape their use (tokens, context windows, sampling, retrieval, fine-tuning) and their risks (ethics, interpretability, hallucination). The field moves quickly, so definitions of broad terms are stable while specific model names, sizes and benchmarks age within months. An entry should be read as the concept, not as a snapshot of the state of the art.
Key ideas
- Token
- The unit of text a language model reads and writes, often a word fragment. Costs and context limits are measured in tokens, not characters.
- Context window
- The maximum number of tokens a model can consider at once, covering the prompt and the response together.
- Training vs inference
- Training adjusts model parameters from data; inference uses the trained model to produce outputs. Their costs and tools differ.
- Overfitting and regularization
- Overfitting is memorizing training data at the expense of generalization; regularization techniques reduce it.
- Embedding
- A vector representation of text, images or other data in which similar items are close together.
- Retrieval-augmented generation
- Supplying a model with retrieved documents at query time so answers can use current or private information.
- Evaluation metric
- A number that summarizes performance on a task. Different metrics reward different behavior, so the choice shapes what gets optimized.
Common pitfalls
- Treating model output as verified fact. Language models can produce fluent, wrong statements (hallucination).
- Choosing a metric that does not match the cost of errors; accuracy is misleading on imbalanced data.
- Quoting model sizes, context lengths or benchmark scores as current without a date.
- Confusing correlation in training data with understanding, or assuming interpretability methods explain a model exactly.
Using it from an agent
- Distinguish a concept (attention) from a named product or model, and date any claim about a specific model.
- Ground factual answers in retrieved sources and cite them; verify numbers against reference data before stating them.
- When explaining a metric, state its formula and what kind of error it ignores.
Start with: Large Language Model Glossary · Machine Learning Evaluation Metrics
Datasets in AI & ML Terms
-
AI Ethics & Interpretability
Key concepts in responsible AI: fairness, bias detection, explainability (XAI), model cards, and regulatory compliance (EU AI Act).
13 entries -
Attention Mechanisms in Deep Learning
Attention architectures: self-attention, multi-head attention, cross-attention, scaled dot-product, sparse attention, and efficient attention variants (Linformer, Performer, FlashAttention).
8 entries -
Deep Learning Optimizers
Optimization algorithms for training neural networks: SGD, Momentum, NAG, Adam, AdamW, RMSprop, AdaGrad, and learning rate scheduling strategies.
8 entries -
LLM Evaluation Failure Modes
Evaluation mistakes that inflate confidence in language-model and agent changes, with corrective checks.
10 entries -
Large Language Model Glossary
Key terms and concepts in LLM technology: tokens, context windows, temperature, RAG, fine-tuning, quantization, KV cache, and emergence.
41 entries -
Machine Learning Evaluation Metrics
Common ML evaluation metrics: Accuracy, Precision, Recall, F1 Score, ROC-AUC, MAE, MSE, and R-squared with formulas and best-use scenarios.
21 entries -
Machine Learning Model Architectures
Overview of major ML model families: feedforward neural networks, CNNs, RNNs/LSTMs, Transformers, GANs, and when to use each.
7 entries -
Machine Learning Training Algorithms
Core ML training algorithms: gradient descent variants, backpropagation, optimization methods (SGD, Adam, RMSprop), and when to use each.
13 entries -
Neural Network Architectures
Common neural network architectures: CNN, RNN, LSTM, Transformer, GAN, and Diffusion models with their primary use cases and key features.
12 entries -
Regularization Techniques in Machine Learning
Techniques to prevent overfitting: L1/L2 regularization, dropout, batch normalization, early stopping, data augmentation, label smoothing, and mixup.
8 entries -
Retrieval Evaluation Metrics for RAG
Retrieval metrics and diagnostic slices for evaluating the evidence selected before language-model generation.
10 entries -
Transfer Learning and Fine-Tuning
Transfer learning strategies: feature extraction, fine-tuning, domain adaptation, multi-task learning, prompt tuning, LoRA, and pretraining paradigms.
8 entries
Fetch this category
One request returns the category with all of its dataset summaries. Append
?format=yaml or ?format=toon for the other encodings.
GET https://yjtoon.yunusemrevurgun.com/api/category/ai-ml-terms?format=json