Weights & Biases
by CoreWeave
ML experiment tracking, model registry and LLM tracing (Weave) platform, part of CoreWeave Forge, read through its Public API.
- Category
- Cloud & Data Platforms
- Type
- MLOps
- Interface
- SDK
- Built with
- Connect Anything
- Credentials
- API key
02Data
What you can sync.
- Runs
- Run metrics
- Artifacts
- Sweeps
- Reports
Load Weights & Biases runs into your Lake and model it in the Warehouse with the rest of your stack.
Need more? Name any other Weights & Biases data the API returns, and the AI adds it to the connector.
03Credentials
Credentials you will need.
- An API key
Weights & Biases uses an API key. You add the credentials once, when the AI asks for them, and it tells you where to create them in CoreWeave if you need help.
04How it works
How it works in DataMind.
-
Pick Weights & Biases.
Open Lake, pick Weights & Biases in the source picker and name the connection. Connect Anything opens with a ready prompt that points the AI at the Weights & Biases API docs.
-
Answer a few questions.
The AI reads the Weights & Biases documentation, then asks which credentials you will use, which data to sync (for example runs and run metrics), whether you want a one-time load or a recurring sync, and which accounts or filters apply.
-
Your data lands in Lake.
The connector agent writes a Python connector for Weights & Biases, tests it against the live API and shows you sample rows. Then the data loads into your Lake on your schedule, ready for the Warehouse, DataPrism BI and AI Minds.
