Weights & Biases (W&B), now part of CoreWeave's Forge platform, is a tool for tracking machine learning experiments and for evaluating AI applications.
In plain words
Training a model means many runs with different settings. With a few lines of code, W&B records each run's settings, such as learning rate, and its metrics, such as loss and accuracy, then shows them on dashboards so you can compare runs. A companion tool, W&B Weave, evaluates apps and agents built on language models.
A quick example
Imagine a student training an image classifier ten times with different learning rates. Each run logs its accuracy to W&B, and a single chart shows which setting learned fastest, instead of the student copying numbers into a spreadsheet.
Where you'll meet it
- Comparing training runs in research and industry.
- Evaluating and tracing LLM apps.
Related
Sources
- DocsForge Home, CoreWeave Docs · read 14 Oct 2026
- DocsExperiments overview, Weights & Biases · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 14 Oct 2026. Spotted an error? Report it.