seeds/42/summary.json

{
  "ai_token": "/workspace/expanded_raid_ai_20260726/outputs/gemma_ai_pairwise_finecurve5k_lr1e4_seed42/full/step_00075/tokens/ai_token.pt",
  "beemo": {
    "accuracy_at_zero": 0.7059639389736477,
    "auroc": 0.754864147563069,
    "auroc_ci95": [
      0.7407155051042317,
      0.7685814134621077
    ],
    "mean_ai_score": 8.748883680532197,
    "mean_human_score": 0.3008289260551216,
    "pairs": 2163,
    "texts": 4326
  },
  "eval_only": true,
  "human_token": "/workspace/rebuttal_apo_20260724/outputs/rebuttal_text_multiseed_20260726/host_b_complete/gemma4_human_steps900_20260726/tokens/human_token.pt",
  "raid_standard_test": {
    "accuracy_at_zero": 0.848,
    "auroc": 0.9641748888888888,
    "auroc_ci95": [
      0.9597294527777779,
      0.9685374472222222
    ],
    "mean_ai_score": 6.8369182267834745,
    "mean_human_score": -12.562866653303304,
    "pairs": 3000,
    "texts": 6000
  }
}