
Weights & Biases
The ML experiment-tracking standard, now with Weave for LLM tracing and evals.
Acquired — still operating
acquired by CoreWeave
About
Weights & Biases is the long-standing standard for ML experiment tracking: metrics, hyperparameters, artifacts, and model registry, with the wandb/wandb client MIT-licensed (~11k stars). Weave (wandb/weave) extends it to LLM tracing and evaluations.
Closed SaaS platform (acquired by CoreWeave). Free tier available. If you already track ML experiments in W&B, Weave is the natural LLM extension; greenfield LLM-observability shoppers usually pick Langfuse or LangSmith.
What it can do
- Experiment tracking: metrics, params, artifacts
- Model registry and versioning
- Weave: LLM tracing and evaluations
- Sweeps for hyperparameter search
- Reports and collaboration
Setup
- Sign up at wandb.ai.
pip install wandband runwandb login.- Wrap training runs with
wandb.init(). - For LLM tracing:
pip install weaveand use@weave.op()decorators.
Prerequisites
W&B account (free tier); API key.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier). The established ML-tracking standard (~11k stars, MIT client, active), official site live, Weave covers the LLM surface. Docked because the platform is closed SaaS (now under CoreWeave) and the issue count is very high — though that reflects scale. Verified tier.
★ 11,276 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 7.7/10
- Maintenance · recency, releases, issue load
- 8.2/10
- Openness · license + docs
- 9.7/10
- Manual review · sanity read + red-flag check
- 9.0/10
- · 11,276 GitHub stars (log-scaled vs catalog peers)
- · last push 1d ago; 18 releases/12mo; 1007 open issues
- · manual review: the ML experiment-tracking standard; Weave adds LLM tracing; very high issue count reflects scale; license: MIT (client); platform closed SaaS
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.