Improve · Datasets
Know which data a research workflow is allowed to use
Dataset choice affects what a team can reproduce, evaluate and legitimately reuse. SAVRN Cloud’s planned dataset workflow brings purpose, provenance and rights into the same discussion as model development. Its public preview uses small browser-local examples. Connected dataset governance and storage are Coming soon.
Begin with existing source discovery
The SAVRN dataset catalog is an existing discovery destination at /datasets. Use it to find potential sources and review the information available for each entry. Discovery does not grant usage rights, authorize institutional processing or import a dataset into Cloud. Source suitability remains a separate decision for the proposed research task.
Keep training and evaluation purposes distinct
The preview lets you add a small sample dataset with a description, a purpose and a rights note. A training-example collection can enter the adapter workflow; a frozen evaluation set can enter a test. These purpose labels illustrate intended controls. They do not verify legal rights, content independence or the quality of the supplied examples.
Make a revision explainable
The planned connected record would identify the source, authorized uses, transformations and version used by a particular run. Changes to cleaning, selection or labels can alter a result even if the dataset’s display name remains the same. Research reproducibility depends on preserving those distinctions and documenting exclusions, rather than treating a collection as an unchanging folder.
Match handling to the approved use
Institutional data may need specific storage, retention, access and export controls. Those requirements need assessment before ingestion and must remain enforceable throughout execution and artifact handling. The preview provides no approved research repository or automatic policy enforcement. Use public or invented sample lines, and retain simulation labels when exporting a demonstration collection.
Service maturity
Explore the workflow today
Create and export small local example collections with training or evaluation purposes.
Before connected service access
Connected dataset services require approved storage, enforceable access, rights review and reliable provenance and revision records.
What this makes possible
The intended outcome is a traceable relationship between a dataset revision, its permitted purpose and the research results produced from it.
Build toward the work that matters.
Tell SAVRN what your institution needs to run, who reviews the results and where its data must stay. That workload defines the next service to qualify.
Discuss an AI project