{
  "slug": "domain-mixture",
  "domain": "AI foundations",
  "title": "Predicting the effect of a data mixture.",
  "description": "A fitted scaling-law program that predicts held-out loss across training-domain mixtures.",
  "metric": "Held-out R²",
  "unit": "",
  "note": "The figure shows predictions on the frozen held-out data from the paper’s replay.",
  "task": "Domain Mixture Scaling",
  "score": 0.9970620845239324,
  "reference": 0.9969219209967796,
  "direction": "max",
  "outcome": "Improved",
  "model": "GPT-5.6-Luna",
  "budget": 8,
  "seed": 43,
  "best_iteration": 96,
  "last_iteration": 100,
  "cost_usd": 15.83991904,
  "program_id": "3425689e-66d0-4168-87a5-8544bb7bf7f5",
  "source_sha256": "6b2f917e8dfee7016ba04d4cd9f6b98d20bb6006950eeb5873ce6b1f56e2f48a",
  "source": "static/programs/domain_mixture.py",
  "language": "Python",
  "lines": 104,
  "history": {
    "points": [
      {
        "iteration": 0,
        "candidate_score": 0.9043309853016344,
        "best_score": 0.9043309853016344,
        "gate": "unrecorded"
      },
      {
        "iteration": 1,
        "candidate_score": 0.980030269471992,
        "best_score": 0.980030269471992,
        "gate": "noop"
      },
      {
        "iteration": 2,
        "candidate_score": 0.9800974597813576,
        "best_score": 0.9800974597813576,
        "gate": "noop"
      },
      {
        "iteration": 3,
        "candidate_score": 0.9933587147731237,
        "best_score": 0.9933587147731237,
        "gate": "noop"
      },
      {
        "iteration": 4,
        "candidate_score": 0.9933587147731237,
        "best_score": 0.9933587147731237,
        "gate": "noop"
      },
      {
        "iteration": 5,
        "candidate_score": 0.9933587147731237,
        "best_score": 0.9933587147731237,
        "gate": "noop"
      },
      {
        "iteration": 6,
        "candidate_score": 0.9933587147731237,
        "best_score": 0.9933587147731237,
        "gate": "noop"
      },
      {
        "iteration": 7,
        "candidate_score": 0.9988519578112588,
        "best_score": 0.9988519578112588,
        "gate": "retrieve"
      },
      {
        "iteration": 8,
        "candidate_score": 0.9988519578112588,
        "best_score": 0.9988519578112588,
        "gate": "lookup"
      },
      {
        "iteration": 9,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "noop"
      },
      {
        "iteration": 10,
        "candidate_score": 0.998851188812546,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 11,
        "candidate_score": 0.998851504983809,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 12,
        "candidate_score": 0.9967862982660229,
        "best_score": 0.9988649080773501,
        "gate": "noop"
      },
      {
        "iteration": 13,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "noop"
      },
      {
        "iteration": 14,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 15,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "noop"
      },
      {
        "iteration": 16,
        "candidate_score": 0.998851504983809,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 17,
        "candidate_score": 0.9988520836546055,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 18,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "noop"
      },
      {
        "iteration": 19,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "noop"
      },
      {
        "iteration": 20,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 21,
        "candidate_score": 0.998851504983809,
        "best_score": 0.9988649080773501,
        "gate": "noop"
      },
      {
        "iteration": 22,
        "candidate_score": 0.9986967228524068,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 23,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 24,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 25,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 26,
        "candidate_score": 0.998851504983809,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 27,
        "candidate_score": 0.9988520836546056,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 28,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 29,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 30,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 31,
        "candidate_score": 0.9988519578112588,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 32,
        "candidate_score": 0.9988520836546056,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 33,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 34,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 35,
        "candidate_score": 0.9988649073486146,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 36,
        "candidate_score": 0.9988517740361593,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 37,
        "candidate_score": 0.9988523085797234,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 38,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 39,
        "candidate_score": 0.9988645155131032,
        "best_score": 0.9988649080773501,
        "gate": "noop"
      },
      {
        "iteration": 40,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 41,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 42,
        "candidate_score": 0.9988523085797234,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 43,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 44,
        "candidate_score": 0.9988649070043193,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 45,
        "candidate_score": 0.9988618396122121,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 46,
        "candidate_score": 0.9988523086112088,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 47,
        "candidate_score": 0.9988644286927095,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 48,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 49,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 50,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 51,
        "candidate_score": 0.998861002443168,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 52,
        "candidate_score": 0.9988644286927095,
        "best_score": 0.9988649080773501,
        "gate": "noop"
      },
      {
        "iteration": 53,
        "candidate_score": 0.9988520733632126,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 54,
        "candidate_score": 0.9988647805604356,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 55,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 56,
        "candidate_score": 0.9988523086110701,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 57,
        "candidate_score": 0.9988647805604356,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 58,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 59,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "lookup"
      },
      {
        "iteration": 60,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 61,
        "candidate_score": 0.9988523086284751,
        "best_score": 0.9988649080773501,
        "gate": "retrieve"
      },
      {
        "iteration": 62,
        "candidate_score": 0.9988649973477988,
        "best_score": 0.9988649973477988,
        "gate": "retrieve"
      },
      {
        "iteration": 63,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649973477988,
        "gate": "lookup"
      },
      {
        "iteration": 64,
        "candidate_score": 0.9988649979738201,
        "best_score": 0.9988649979738201,
        "gate": "lookup"
      },
      {
        "iteration": 65,
        "candidate_score": 0.9988649979738201,
        "best_score": 0.9988649979738201,
        "gate": "lookup"
      },
      {
        "iteration": 66,
        "candidate_score": 0.9988649979738204,
        "best_score": 0.9988649979738204,
        "gate": "lookup"
      },
      {
        "iteration": 67,
        "candidate_score": 0.9988649973477988,
        "best_score": 0.9988649979738204,
        "gate": "lookup"
      },
      {
        "iteration": 68,
        "candidate_score": 0.9988649080773501,
        "best_score": 0.9988649979738204,
        "gate": "lookup"
      },
      {
        "iteration": 69,
        "candidate_score": 0.9988649080665574,
        "best_score": 0.9988649979738204,
        "gate": "lookup"
      },
      {
        "iteration": 70,
        "candidate_score": 0.9988645125158992,
        "best_score": 0.9988649979738204,
        "gate": "lookup"
      },
      {
        "iteration": 71,
        "candidate_score": 0.9988650358355007,
        "best_score": 0.9988650358355007,
        "gate": "retrieve"
      },
      {
        "iteration": 72,
        "candidate_score": 0.6223906362636485,
        "best_score": 0.9988650358355007,
        "gate": "noop"
      },
      {
        "iteration": 73,
        "candidate_score": 0.9988649979738224,
        "best_score": 0.9988650358355007,
        "gate": "lookup"
      },
      {
        "iteration": 74,
        "candidate_score": 0.9988649979738201,
        "best_score": 0.9988650358355007,
        "gate": "retrieve"
      },
      {
        "iteration": 75,
        "candidate_score": 0.9988650358355007,
        "best_score": 0.9988650358355007,
        "gate": "retrieve"
      },
      {
        "iteration": 76,
        "candidate_score": 0.9988650361486885,
        "best_score": 0.9988650361486885,
        "gate": "lookup"
      },
      {
        "iteration": 77,
        "candidate_score": 0.9988650361518671,
        "best_score": 0.9988650361518671,
        "gate": "lookup"
      },
      {
        "iteration": 78,
        "candidate_score": 0.9988649979738224,
        "best_score": 0.9988650361518671,
        "gate": "lookup"
      },
      {
        "iteration": 79,
        "candidate_score": 0.999255008982568,
        "best_score": 0.999255008982568,
        "gate": "lookup"
      },
      {
        "iteration": 80,
        "candidate_score": 0.9938205927740463,
        "best_score": 0.999255008982568,
        "gate": "lookup"
      },
      {
        "iteration": 81,
        "candidate_score": 0.9977947957345099,
        "best_score": 0.999255008982568,
        "gate": "noop"
      },
      {
        "iteration": 82,
        "candidate_score": 0.9988650361518565,
        "best_score": 0.999255008982568,
        "gate": "lookup"
      },
      {
        "iteration": 83,
        "candidate_score": 0.9988642513445561,
        "best_score": 0.999255008982568,
        "gate": "lookup"
      },
      {
        "iteration": 84,
        "candidate_score": 0.999255008982568,
        "best_score": 0.999255008982568,
        "gate": "lookup"
      },
      {
        "iteration": 85,
        "candidate_score": 0.9992584861983906,
        "best_score": 0.9992584861983906,
        "gate": "lookup"
      },
      {
        "iteration": 86,
        "candidate_score": -52.351125447325344,
        "best_score": 0.9992584861983906,
        "gate": "lookup"
      },
      {
        "iteration": 87,
        "candidate_score": 0.9938205927740452,
        "best_score": 0.9992584861983906,
        "gate": "lookup"
      },
      {
        "iteration": 88,
        "candidate_score": 0.9938205927740409,
        "best_score": 0.9992584861983906,
        "gate": "lookup"
      },
      {
        "iteration": 89,
        "candidate_score": 0.9988606885199097,
        "best_score": 0.9992584861983906,
        "gate": "lookup"
      },
      {
        "iteration": 90,
        "candidate_score": 0.9988650358355007,
        "best_score": 0.9992584861983906,
        "gate": "lookup"
      },
      {
        "iteration": 91,
        "candidate_score": 0.9992584861983906,
        "best_score": 0.9992584861983906,
        "gate": "lookup"
      },
      {
        "iteration": 92,
        "candidate_score": 0.9988650361518671,
        "best_score": 0.9992584861983906,
        "gate": "noop"
      },
      {
        "iteration": 93,
        "candidate_score": 0.9938205927740463,
        "best_score": 0.9992584861983906,
        "gate": "lookup"
      },
      {
        "iteration": 94,
        "candidate_score": -15158.876004014124,
        "best_score": 0.9992584861983906,
        "gate": "lookup"
      },
      {
        "iteration": 95,
        "candidate_score": 0.9988649979738201,
        "best_score": 0.9992584861983906,
        "gate": "lookup"
      },
      {
        "iteration": 96,
        "candidate_score": 0.9992584861983913,
        "best_score": 0.9992584861983913,
        "gate": "lookup"
      },
      {
        "iteration": 97,
        "candidate_score": 0.9938205927740406,
        "best_score": 0.9992584861983913,
        "gate": "lookup"
      },
      {
        "iteration": 98,
        "candidate_score": 0.9938205927740459,
        "best_score": 0.9992584861983913,
        "gate": "lookup"
      },
      {
        "iteration": 99,
        "candidate_score": 0.993820592774039,
        "best_score": 0.9992584861983913,
        "gate": "lookup"
      },
      {
        "iteration": 100,
        "candidate_score": -2331.1104390834935,
        "best_score": 0.9992584861983913,
        "gate": "lookup"
      }
    ],
    "trace_sha256": "6e8eff996557a5c2f64a89916b45c1921eae10c1c907ec34f64ca47c1c987b89",
    "score_label": "Search-time evaluator score ↑",
    "note": "Best-so-far envelope of recorded, non-migrant programs. If multiple programs share an iteration, the highest recorded score is used. Missing iterations are not invented."
  },
  "run_fingerprint": "d21b82ae6b096be388bfaa2057add3842521af9dc6a5aa6f9c37124b533ea17b",
  "artifact": {
    "views": [
      {
        "image": "static/images/programs/domain-mixture-comparison.webp",
        "thumbnail": "static/images/programs/domain-mixture-comparison-thumb.webp",
        "pdf": "static/figures/programs/domain-mixture-comparison.pdf",
        "paper_pdf": "static/figures/programs/paper-ai.pdf",
        "provenance": {
          "source": "figures/supp_figures/sota_objects_ai.pdf",
          "source_sha256": "bbb0bd9bdd63daf56779fa78042a0044a0078f70e5ac1b2b3717f995316ee430",
          "clip_points": [
            0.0,
            119.0,
            233.0,
            267.8399963378906
          ],
          "cropped": true,
          "pdf_sha256": "4f465bebdb5251455fa19ba64c43fe934ea753715bf79fc335bcd93f2a7c3fd1",
          "image_sha256": "32b69bccbc8c2d0ea8406bb37b29aad95856c1a22d192dcb1c57cf845a60f9c0",
          "thumbnail_sha256": "2e90845482e3fec506852592541f15af40d956c11327f53d61dbe9d010053dbf"
        },
        "label": "Predictions",
        "caption": "Predicted versus actual held-out loss for 24 mixtures × 5 domains, with the discovered laws and both methods’ held-out R²."
      },
      {
        "image": "static/images/programs/scaling-held-out-error.webp",
        "thumbnail": "static/images/programs/scaling-held-out-error-thumb.webp",
        "pdf": "static/figures/programs/scaling-held-out-error.pdf",
        "paper_pdf": "static/figures/programs/paper-ai.pdf",
        "provenance": {
          "source": "figures/supp_figures/sota_objects_ai.pdf",
          "source_sha256": "bbb0bd9bdd63daf56779fa78042a0044a0078f70e5ac1b2b3717f995316ee430",
          "clip_points": [
            241.0,
            119.4000015258789,
            396.0,
            246.0
          ],
          "cropped": true,
          "pdf_sha256": "88df75293787c1b7a9f9fe8ef971dc2079c3abbde2901bb2d5dc6925d2c094b4",
          "image_sha256": "f5668d60aa9cd53b5637f5bcc291e20149aff17d04ebd03958a82dc07dcae33c",
          "thumbnail_sha256": "6f32f4909c5aac07b650b99a254ac2adeeec5fb371744020d786404a0d8d8f11"
        },
        "label": "Held-out error",
        "caption": "SimpleTES (purple) and EvoDuet (blue): held-out 1 − R² on a log scale for both scaling benchmarks. Lower is better."
      }
    ],
    "kind": "views",
    "data": {
      "replay_score": 0.9970620845239324,
      "reported_score": 0.9970620845239324,
      "source_matches_selected": true
    }
  },
  "related_trajectories": []
}
