April 2026 — NeoData 1.2: training clarity and operational visibility
The NeoData 1.2 line is aimed at teams that run training at scale and need clear configuration, repeatable recipes, and better visibility into what the platform is doing.
What's new
- Training and evaluation alignment — Configuration paths that tie training setup to evaluation more explicitly, so benchmarks and checkpoints match what you intended to measure.
- Recipe-oriented training — Support for structured training “recipes” and overview surfaces that make complex setups easier to reuse and review.
- Activity and monitoring — Refinements to how activity is presented in the portal, so operators can follow events without digging through internal logs.
Improvements
- Dataset experience — Fixes and polish around dataset naming and validation early in the 1.2 line, reducing confusion when many datasets are in flight.
- Hosting and access — Streamlined front-door configuration for deployments that use a single domain, simplifying browser and cookie behavior for end users.
- Queries and filters — Tighter behavior for training-related queries (for example foundation trainings), so lists reflect the right jobs and states.
Notes for administrators
- If you rely on schedule policies or automation around training windows, re-check schedules after upgrading to the latest 1.2 patch in your environment.
- No broad breaking changes to public integration contracts are called out for this line; confirm with NeoSpace if you embed the portal in iframes or custom shells.