The eager result of evaluating one target over one dataset.
observations keeps every attempted, failed, interrupted, and skipped case.
summaries keeps raw statistics for each named score. selected_score is
the higher-is-better value that search strategies compare.
Summary
Functions
Starts an evaluation under the caller's Task.Supervisor.
Returns the summary for one named score.
Types
@type t() :: %Tune.Evaluation{ calls: %{optional(atom()) => non_neg_integer()}, complete?: boolean(), eligible?: boolean(), metadata: map(), observations: [Tune.Observation.t()], selected_score: number() | nil, stop_reason: :completed | :case_limit | :model_call_limit | :deadline | :cancelled | :lost_accounting, summaries: %{optional(atom()) => Tune.Score.Summary.t()}, usage: %{ input: non_neg_integer(), output: non_neg_integer(), total: non_neg_integer() } }
Functions
@spec start(GenServer.server(), Tune.t(), :train | :validation | :test) :: Task.t()
Starts an evaluation under the caller's Task.Supervisor.
@spec summary(t(), atom()) :: Tune.Score.Summary.t() | nil
Returns the summary for one named score.