About KidML

What KidML is

KidML is an independent, advertising-supported site built around six free calculators for anyone learning machine learning — a kid writing a first line of code, a student sizing a first dataset, a hobbyist training a first model. It isn’t affiliated with any university, framework, or company; it’s tools and reference guides, nothing more.

The tools

Six calculators, each answering one concrete question:

• ML Algorithm Selector — four quick questions about your task and data produce a beginner-friendly shortlist of algorithms to try first.
• Dataset Size Calculator — estimates memory from sample count, feature count, and numeric precision, and splits samples into train, validation, and test sets.
• Model Accuracy Predictor — turns a confusion matrix into accuracy, precision, recall, F1 score, and specificity.
• GPU Memory Calculator — estimates training VRAM from parameter count, batch size, precision, and optimizer, broken into parameters, gradients, optimizer state, and activations.
• Neural Network Calculator — counts a fully-connected network’s weights, biases, and total parameters, layer by layer.
• Training Time Estimator — projects wall-clock training time from dataset size, epochs, batch size, and hardware throughput.

How they work

None of the six run an actual model. Each takes the numbers you type in — features, batch size, layer sizes, a confusion matrix — and applies a fixed formula, showing its steps rather than hiding them behind a single answer. No training happens in your browser, and nothing you enter is sent anywhere.

Where the numbers come from

Parameter counts use the standard weights-plus-biases sum for dense layers. GPU memory follows the usual parameters, gradients, optimizer-state, and activations breakdown. Accuracy, precision, recall, F1, and specificity come directly out of the confusion matrix you provide. Dataset memory is sample count times feature count times bytes per value. If a result doesn’t match what you see in your own framework, that’s worth a note through the contact page.

For younger learners

The wording is deliberately plain — short questions, no assumed background — because a curious ten-year-old and a graduate student usually start with the same questions about datasets and models. That said, these are calculators, not a coding environment or a curriculum; pair them with an actual course, notebook, or class when you’re ready to build something.

The museum

The Machines & Minds Museum is a separate, browsable timeline covering computing history, the rise of robots and AI, and the scientists behind the field. Every entry, image, and fact in it is drawn from Wikipedia and Wikimedia Commons, not written by KidML.

Guides & blog

The blog and the learn sectioncover the ideas behind the numbers — what a confusion matrix actually tells you, why batch size changes memory use, and similar ground the calculators don’t have room to explain.

Straight talk about ads & advice

KidML runs on display ads; that’s what keeps six calculators free with no account and no paywall. The site has no connection to any ML framework, platform, or company its guides happen to mention. Everything here is general educational content, not professional data-science or engineering advice — validate any real project against your own data, hardware, and framework before relying on it.

Questions?

Get in touchif a tool gives you a result that looks wrong, or a formula doesn’t match what you expected.