Polyaxon experimentation is a set of tools for:
- Processing data and Training ML models
- Running Notebooks and Tensorboards, and launching apps and dashboards
- Running distributed jobs using cloud-native operators
- Tracking metrics and artifacts
- Driving analyses and visualizing results
Polyaxon experimentation tools are:
- Reproducible: Code, environment, and specification are versioned, dependencies are containerized, and every execution is tracked and reproducible.
- Serverless and Kubernetes native so they benefit from all the benefits containerization provides: portability, scalability, reliability.
- Massive scale: Fully distributed, several scheduling strategies, fault-tolerant, scale to multiple nodes, clusters, and hundreds of thousands of concurrent executions.
- Sharable: Any component can be promoted with typed inputs and outputs to be sharable and reusable.