# Experiment Tracking

> Notes on each training run. You can then compare them and redo the best one.

**Field:** Artificial Intelligence  
**Category:** MLOps & Deployment  
**Also known as:** ML Experiment Tracking, Run Tracking

## Think of it like

Think of it like a science notebook. You write down what you tried and what happened, every time.

## Definition

Experiment tracking means saving a record of every training run, so you can compare runs and repeat the best one. Each record holds the settings used, the version of the code and the data, the scores and the files the run made. A dashboard then shows runs side by side, so you can see which change helped. Without it, people forget which settings gave the best model. It is often the first MLOps tool a team adds.

## Example

Emma tries 40 versions of a model in a week. Her tracking tool shows that run 27 scored best, with a smaller learning rate, so she knows how to rebuild it.

## Why it matters

It turns guesswork into a record. You can always say how a model was made and why it was picked.

## Common mistake

Tracking scores but not the data version. If the data changed between runs, the scores can't be fairly compared.

## Three principles

- **Every run saved.** Settings and scores.
- **Code and data versions.** Both recorded.
- **Compare side by side.** Pick the best.

## Related terms

- [ML Pipeline](https://learn.benibix.com/term/ai/ml-pipeline)
- [Hyperparameter](https://learn.benibix.com/term/ai/hyperparameter)
- [Model Registry](https://learn.benibix.com/term/ai/model-registry)
- [Version Control](https://learn.benibix.com/term/development/version-control)
- [MLflow](https://learn.benibix.com/term/ai/mlflow)
- [Weights & Biases (W&B)](https://learn.benibix.com/term/ai/weights-and-biases)

## Told apart from

- [Experiment Tracking vs Model Registry](https://learn.benibix.com/compare/experiment-tracking-vs-model-registry): People think the model registry is where training runs are logged, and it is not. Experiment tracking logs every run, good or bad, with its settings, data and scores. The registry keeps only the models someone chose to keep, with a stage such as testing or live. Each version links back to its run. Track to find the best model, and register it to ship it.

![A table of training runs. Each row shows the settings used, the data version and the score. The best row is marked.](https://learn.benibix.com/diagrams/mo-experiment-tracking.svg)

---

Source: https://learn.benibix.com/term/ai/experiment-tracking
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-17
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
