ConceptAI & ML
Hyperparameters
Choices about the algorithm that are set rather than learned from training data.
Know first
Named as a prerequisite by the lectures that define this. The number is how many say so. A grey one has no page yet.
Machine learning algorithms1Neural network architecture1Neural networks1Parameter1Reinforcement learning algorithms1Smoothing1Training data1Training process1
Opens up
Concepts that name this one as a prerequisite.
Lecturers
4Each in their own terms, from the easiest level.
Fei-Fei Li, Justin Johnson & Serena Yeung
Andrej Karpathy
A design choice determined by the model creator, such as the size of the hidden layer.Pieter Abbeel
Settings like the amount or type of smoothing, such as k or alpha, that are not learned directly from training data parameters.Matthew Vandergrift
Variables which control the learning process of an agent that must be manually set by the practitioner before deploying the agent.
Lectures that assume it
11- Training Neural Networks II
- CNN Architectures
- Training Neural Networks I
- The Alberta Plan for AI Research: Continual Reinforcement Learning
- Marin: An Open Lab For Frontier AI
- Demystifying Mixtral of Experts
- Approximation Theory
- Representation Learning: Theory
- Challenges and Open Problems in Deep Reinforcement Learning (CS 285, Lecture 23, Part 1)
- Stable Estimators for Fast Private Statistics
- Scan at the Speed of Light
Connections
24The lectures that define it or assume it, and the concepts next to it.
Concepts8
- Neural network architecture
- Neural networks
- Parameter
- Architecture selection for MoE
- Automatic tuning
- Backtracking
Show 2 moreShow fewer
Lectures16
- Building makemore Part 2: Multi-Layer Perceptron Language Models (Bengio et al. 2003)
- Detection and Segmentation
- Image Classification
- Machine Learning: Naive Bayes
- Online Hyperparameter Tuning as an Empirical Approach for Reinforcement Learning
- Training Neural Networks II
Show 10 moreShow fewer
- CNN Architectures
- Training Neural Networks I
- The Alberta Plan for AI Research: Continual Reinforcement Learning
- Marin: An Open Lab For Frontier AI
- Demystifying Mixtral of Experts
- Approximation Theory
- Representation Learning: Theory
- Challenges and Open Problems in Deep Reinforcement Learning (CS 285, Lecture 23, Part 1)
- Stable Estimators for Fast Private Statistics
- Scan at the Speed of Light