Dataset strategy
Data inventory, source rights, provenance records, stewardship roles, quality standards and collection plan.
Service 03
AI models by and for each culture.
Palace Ring coordinates the data, evaluation, training, deployment and transfer work required for models shaped by a participating institution’s languages, knowledge, standards and cultural heritage.
Model development starts with data rights and evaluation criteria.
Programs may cover dataset development, evaluation, fine-tuning, deployment or foundation-model training on institution-controlled infrastructure.
The program specification records language coverage, source rights, provenance, quality standards, security requirements, cultural review and acceptance tests.
Adaptation and evaluation
Base-model decision, fine-tuning assets, evaluation suite, acceptance thresholds and deployment controls.
Foundation models
Training architecture, compute plan, development records, model documentation and operating procedures.
Ownership
Ownership terms are fixed before development begins.
Contract schedules identify data custody, ownership or license rights for training assets and weights, deployment authority, documentation and transfer rights.
Institutional ownership applies only where weights and related assets are created for a participating institution or expressly transferred under contract.