Artificial intelligence: development, training and integration · ENGLISH
Service page for projects serving Cuenca. Decide whether fine-tuning fits the task and define evaluation examples before tuning. In Cuenca, grouping content by need makes schedules, catalogues or services easier to navigate.
01 · CONTEXT
In Cuenca, grouping content by need makes schedules, catalogues or services easier to navigate.
02 · PRACTICAL BRIEF
Models, integration and evaluation
Define the task, representative examples, permitted data, evaluation goal and runtime environment. Leave sensitive data out of the initial brief. Before finalising the brief for Cuenca, confirm with the people involved: In Cuenca, grouping content by need makes schedules, catalogues or services easier to navigate.
An example schema, training format, configuration and base model version.
Evaluate held-out data and check for regressions on tasks the model already handled.
03 · METHOD
Models, integration and evaluation
04 · ANSWERS
Decide whether fine-tuning fits the task and define evaluation examples before tuning. An example schema, training format, configuration and base model version.
Prepare the task, representative examples, permitted data, privacy constraints and target environment. Avoid sharing sensitive data in an initial enquiry.
Evaluate held-out data and check for regressions on tasks the model already handled.