# Dataset construction (deterministic). Seeds define split membership; keys are written to data/generated/keys (evaluator-only).
synthetic:
  n_projects: 6
  lengths_tokens: [2000, 8000, 16000]
  # seed ranges by split (disjoint; template families identical, entities/numbers differ)
  splits:
    train: {start: 1000, count: 40}      # mapper/calibration/prompt tuning
    dev:   {start: 2000, count: 30}      # method selection, router fitting
    test:  {start: 3000, count: 40}      # locked; first evaluated at Phase 8
  length_assignment: round_robin          # each seed gets one length: 2000, 8000, 16000, 2000, ...
exec:
  splits:
    train: {start: 1, count: 20}
    dev:   {start: 101, count: 30}
    test:  {start: 201, count: 40}
  target_tokens: 2000
public:
  subsets: [qasper, 2wikimqa]
  extension_subsets: [hotpotqa]
  split_rule: "sha256(document text) last hex digit: 0-3 -> dev, 4-f -> test (dev ~25%, test ~75%)"
  max_context_tokens: 16384