Everyone's talking about AI adoption. Almost nobody has the real numbers. Help us change that — and get the full report 👉 Engineers | Leaders

latent patterns pricing login

Logits

The raw, unnormalized scores a model outputs before softmax — one value per class or vocabulary token, not yet probabilities.

Latent Patterns is a new platform that teaches AI concepts to developers — through screencasts, technical deep dives, interactive playgrounds, and hands-on courses.

We haven't launched yet. Sign up below and we'll notify you when we open the doors.

[→] RELATED CONTENT

courses

Understanding microGPT Walk through Karpathy's 200-line GPT implementation — from autograd and tokenization to multi-head attention, training, and text generation.
Understanding Makemore Build character-level language models from scratch — from bigram counting through MLPs, batch normalization, and WaveNet — generating new names with PyTorch.

lessons

What Is a GPT? Understanding GPT / The Big Picture
A Bigram Language Model Understanding GPT / The Bigram Baseline
Layer Normalization and Transformer Blocks Understanding GPT / Building the Transformer
The Neural Bigram Understanding Makemore / Bigrams
The MLP Architecture Understanding Makemore / The MLP Language Model
The MLP and Residual Connections Understanding microGPT / The Transformer Architecture
Loss and Next-Token Prediction Understanding microGPT / Training
Sampling and Temperature Understanding microGPT / Generation

learn more?

Subscribe and we'll send new content to your inbox.

learn

courses glossary blog newsletter knuth checks verify

for

enterprise startups consultants bootcamps trainers university government

company

about principles careers press security privacy terms status

connect

x youtube instagram linkedin bluesky discord

© 2026 Latent Patterns