What is Weights & Biases?
Weights & Biases (W&B) is a platform for developing machine learning models. The CoreWeave documentation describes it as covering experiment tracking, hyperparameter sweeps, artifacts, reports and automations, with W&B Weave alongside for LLM and agent evaluation. Pricing and documentation for W&B now sit under CoreWeave Forge, the CoreWeave product family for developing, evaluating and serving AI models and agents.
In the Forge product list, Weights & Biases Models is described as a way to track, compare and reproduce every experiment. Registry is the system of record for models, agents and datasets, and Notebooks are reactive Python notebooks connected to your runs.
The Forge pricing page lists a Free plan that is free to build with any cloud, model or framework, and a Pro plan for professionals working to optimize AI applications and models. Pro starts at $60 per month billed monthly and has a 30-day free trial. Enterprise is a separate plan, and academic research is free forever.
Key capabilities
- Experiment tracking Track, compare and reproduce every experiment, with sweeps and reports.
- Artifacts and Registry A system of record for models, agents and datasets.
- Weave Evaluation for LLM applications and agents.
- Notebooks Reactive Python notebooks connected to your runs.
- Plans Free, Pro from $60 per month, Enterprise, and free use for academic research.
Where Weights & Biases fits
- Experiment tracking Tracks, visualizes and compares machine learning experiments, with hyperparameter sweeps.
- Artifacts and reports Versions datasets and models as artifacts, and shares results in reports and automations.
- Registry Manages models, agents and datasets across their lifecycle.
- LLM tracing Weave traces LLM applications, with prompt management and cost tracking for production AI apps.
- Evaluation Weave evaluates agents and applications, for example RAG applications scored with LLM judges.
- ARIA research agent An AI research agent that analyzes experiments, explains results and builds visualizations and reports.
- Integrations and hosting Connects to popular ML frameworks and runs in the cloud or on private infrastructure, including self-managed AWS, GCP, Azure or on-premises.
Reviews of Weights & Biases
Be the first to share your experience with Weights & Biases.
Write a review
Loading review form…
