Datasets:
Tasks:
Text Classification
Formats:
csv
Languages:
English
Size:
10K - 100K
Tags:
context-engineering
loop-engineering
harness-engineering
llm-agents
systematic-review
literature-review
License:
Download results.json from thanatchon36/context-loop-engineering: direct link, hf CLI and curl.
- Browser
- Download file 29.3 kB
-
https://huggingface.co/datasets/thanatchon36/context-loop-engineering/resolve/main/results.json
- Command line
-
hf download hf://datasets/thanatchon36/context-loop-engineering/results.json
-
curl -L -o results.json https://huggingface.co/datasets/thanatchon36/context-loop-engineering/resolve/main/results.json
29.3 kB
{
"flow": {
"lists": 5,
"entries": 1891,
"unique": 1492,
"excluded": 97,
"excluded_by_reason": {
"excluded section": 58,
"out of scope": 31,
"model architecture": 8
},
"included": 1395,
"entries_by_list": {
"aiboost_harness": 532,
"deepjlu_graph": 447,
"picrew_harness": 385,
"meirtz_context": 348,
"walkinglabs_harness": 179
},
"multi_list_sources": 180
},
"composition": {
"academic": {
"context": 190,
"evaluation": 92,
"harness": 110,
"loop": 152
},
"grey": {
"context": 39,
"evaluation": 38,
"harness": 146,
"loop": 51
},
"repository": {
"context": 61,
"evaluation": 70,
"harness": 346,
"loop": 100
}
},
"composition_layer_totals": {
"harness": 602,
"loop": 303,
"context": 290,
"evaluation": 200
},
"composition_type_totals": {
"repository": 577,
"academic": 544,
"grey": 274
},
"academic_dated_share": 0.945,
"timeline_academic": {
"context": {
"2020H1": 3,
"2020H2": 2,
"2021H1": 1,
"2021H2": 4,
"2022H1": 2,
"2022H2": 10,
"2023H1": 16,
"2023H2": 28,
"2024H1": 40,
"2024H2": 27,
"2025H1": 23,
"2025H2": 4,
"2026H1": 25,
"2026H2": 4
},
"evaluation": {
"2020H1": 0,
"2020H2": 0,
"2021H1": 0,
"2021H2": 0,
"2022H1": 0,
"2022H2": 0,
"2023H1": 2,
"2023H2": 8,
"2024H1": 14,
"2024H2": 6,
"2025H1": 4,
"2025H2": 4,
"2026H1": 23,
"2026H2": 10
},
"harness": {
"2020H1": 1,
"2020H2": 0,
"2021H1": 0,
"2021H2": 0,
"2022H1": 3,
"2022H2": 1,
"2023H1": 6,
"2023H2": 6,
"2024H1": 12,
"2024H2": 9,
"2025H1": 15,
"2025H2": 6,
"2026H1": 34,
"2026H2": 8
},
"loop": {
"2020H1": 0,
"2020H2": 0,
"2021H1": 0,
"2021H2": 0,
"2022H1": 0,
"2022H2": 0,
"2023H1": 7,
"2023H2": 9,
"2024H1": 9,
"2024H2": 10,
"2025H1": 25,
"2025H2": 17,
"2026H1": 54,
"2026H2": 21
}
},
"timeline_academic_yearly": {
"context": {
"2020": 5,
"2021": 5,
"2022": 12,
"2023": 44,
"2024": 67,
"2025": 27,
"2026": 29
},
"evaluation": {
"2020": 0,
"2021": 0,
"2022": 0,
"2023": 10,
"2024": 20,
"2025": 8,
"2026": 33
},
"harness": {
"2020": 1,
"2021": 0,
"2022": 4,
"2023": 12,
"2024": 21,
"2025": 21,
"2026": 42
},
"loop": {
"2020": 0,
"2021": 0,
"2022": 0,
"2023": 16,
"2024": 19,
"2025": 42,
"2026": 75
}
},
"academic_2026_layer_share": {
"loop": 0.419,
"harness": 0.235,
"evaluation": 0.184,
"context": 0.162
},
"academic_all_layer_share": {
"context": 0.37,
"loop": 0.296,
"harness": 0.196,
"evaluation": 0.138
},
"ops_counts": {
"Compose": 100,
"Select": 425,
"Write": 280,
"Compress": 96,
"Isolate": 109,
"Evolve": 33,
"Guard": 142
},
"levels_counts": {
"L1": 115,
"L2": 199,
"L3": 185,
"L4": 13,
"L5": 344,
"L6": 176,
"L7": 48,
"L8": 94
},
"ops_share_in_context_layer": {
"Compose": 0.186,
"Select": 0.534,
"Write": 0.441,
"Compress": 0.179,
"Isolate": 0.014,
"Evolve": 0.038,
"Guard": 0.021
},
"levels_share_in_loop_layer": {
"L1": 0.069,
"L2": 0.277,
"L3": 0.149,
"L4": 0.043,
"L5": 0.488,
"L6": 0.152,
"L7": 0.158,
"L8": 0.238
},
"contribution_by_layer": {
"benchmark": {
"context": 0.0,
"evaluation": 0.445,
"harness": 0.0,
"loop": 0.0
},
"empirical": {
"context": 0.031,
"evaluation": 0.0,
"harness": 0.017,
"loop": 0.017
},
"method": {
"context": 0.548,
"evaluation": 0.235,
"harness": 0.143,
"loop": 0.465
},
"practice_guide": {
"context": 0.124,
"evaluation": 0.08,
"harness": 0.236,
"loop": 0.162
},
"survey": {
"context": 0.09,
"evaluation": 0.015,
"harness": 0.032,
"loop": 0.026
},
"tool": {
"context": 0.207,
"evaluation": 0.225,
"harness": 0.573,
"loop": 0.33
}
},
"contribution_by_layer_counts": {
"benchmark": {
"context": 0,
"evaluation": 89,
"harness": 0,
"loop": 0
},
"empirical": {
"context": 9,
"evaluation": 0,
"harness": 10,
"loop": 5
},
"method": {
"context": 159,
"evaluation": 47,
"harness": 86,
"loop": 141
},
"practice_guide": {
"context": 36,
"evaluation": 16,
"harness": 142,
"loop": 49
},
"survey": {
"context": 26,
"evaluation": 3,
"harness": 19,
"loop": 8
},
"tool": {
"context": 60,
"evaluation": 45,
"harness": 345,
"loop": 100
}
},
"domain_by_layer": {
"dialogue": {
"context": 0.048,
"evaluation": 0.03,
"harness": 0.023,
"loop": 0.013
},
"general": {
"context": 0.738,
"evaluation": 0.615,
"harness": 0.488,
"loop": 0.601
},
"research": {
"context": 0.034,
"evaluation": 0.02,
"harness": 0.055,
"loop": 0.04
},
"software": {
"context": 0.166,
"evaluation": 0.265,
"harness": 0.387,
"loop": 0.314
},
"web_gui": {
"context": 0.014,
"evaluation": 0.07,
"harness": 0.047,
"loop": 0.033
}
},
"domain_totals": {
"general": 813,
"software": 429,
"research": 59,
"web_gui": 56,
"dialogue": 38
},
"evaluative_share_by_layer": {
"context": 0.031,
"evaluation": 0.445,
"harness": 0.017,
"loop": 0.017
},
"practice_share_by_layer": {
"context": 0.345,
"evaluation": 0.54,
"harness": 0.817,
"loop": 0.498
},
"first_academic_use": {
"context engineering": {
"first_arxiv_yymm": "2510",
"n_titles_or_notes": 2
},
"harness engineering": {
"first_arxiv_yymm": "2604",
"n_titles_or_notes": 5
},
"loop engineering": {
"first_arxiv_yymm": "2607",
"n_titles_or_notes": 1
}
},
"coupling": {
"n_sources": 1395,
"pairs_tested": 56,
"pairs_significant_holm": 6,
"significant_pairs": [
{
"level_name": "Improvement",
"op": "Evolve",
"both": 20,
"lift": 8.994,
"p_holm": 0.0
},
{
"level_name": "Scheduled",
"op": "Isolate",
"both": 11,
"lift": 2.933,
"p_holm": 0.0392
},
{
"level_name": "Session",
"op": "Isolate",
"both": 35,
"lift": 2.545,
"p_holm": 0.0
},
{
"level_name": "Action",
"op": "Compress",
"both": 20,
"lift": 2.527,
"p_holm": 0.0024
},
{
"level_name": "Session",
"op": "Write",
"both": 65,
"lift": 1.84,
"p_holm": 0.0
},
{
"level_name": "Orchestration",
"op": "Isolate",
"both": 48,
"lift": 1.786,
"p_holm": 0.0002
}
],
"draft_pairs": 19,
"draft_pairs_lift_gt1": 12,
"draft_pairs_significant": 4,
"nondraft_significant": 2,
"mean_lift_draft": 1.812,
"mean_lift_nondraft": 0.973
},
"github_monthly": {
"2025-01": {
"context engineering": 3,
"harness engineering": 0,
"loop engineering": 0,
"prompt engineering": 177
},
"2025-02": {
"context engineering": 2,
"harness engineering": 1,
"loop engineering": 0,
"prompt engineering": 232
},
"2025-03": {
"context engineering": 2,
"harness engineering": 0,
"loop engineering": 1,
"prompt engineering": 299
},
"2025-04": {
"context engineering": 3,
"harness engineering": 2,
"loop engineering": 0,
"prompt engineering": 437
},
"2025-05": {
"context engineering": 4,
"harness engineering": 0,
"loop engineering": 1,
"prompt engineering": 387
},
"2025-06": {
"context engineering": 11,
"harness engineering": 1,
"loop engineering": 0,
"prompt engineering": 497
},
"2025-07": {
"context engineering": 184,
"harness engineering": 2,
"loop engineering": 1,
"prompt engineering": 659
},
"2025-08": {
"context engineering": 126,
"harness engineering": 0,
"loop engineering": 0,
"prompt engineering": 610
},
"2025-09": {
"context engineering": 76,
"harness engineering": 1,
"loop engineering": 0,
"prompt engineering": 554
},
"2025-10": {
"context engineering": 147,
"harness engineering": 0,
"loop engineering": 0,
"prompt engineering": 624
},
"2025-11": {
"context engineering": 111,
"harness engineering": 0,
"loop engineering": 1,
"prompt engineering": 548
},
"2025-12": {
"context engineering": 125,
"harness engineering": 2,
"loop engineering": 1,
"prompt engineering": 735
},
"2026-01": {
"context engineering": 132,
"harness engineering": 4,
"loop engineering": 2,
"prompt engineering": 726
},
"2026-02": {
"context engineering": 133,
"harness engineering": 25,
"loop engineering": 5,
"prompt engineering": 826
},
"2026-03": {
"context engineering": 207,
"harness engineering": 242,
"loop engineering": 3,
"prompt engineering": 1330
},
"2026-04": {
"context engineering": 190,
"harness engineering": 408,
"loop engineering": 8,
"prompt engineering": 1267
},
"2026-05": {
"context engineering": 191,
"harness engineering": 321,
"loop engineering": 14,
"prompt engineering": 1431
},
"2026-06": {
"context engineering": 168,
"harness engineering": 255,
"loop engineering": 375,
"prompt engineering": 1834
},
"2026-07": {
"context engineering": 181,
"harness engineering": 261,
"loop engineering": 426,
"prompt engineering": 1898
},
"2026-08": {
"context engineering": 208,
"harness engineering": 204,
"loop engineering": 304,
"prompt engineering": 1355
},
"2026-09": {
"context engineering": 199,
"harness engineering": 360,
"loop engineering": 143,
"prompt engineering": 1470
},
"2026-10": {
"context engineering": 189,
"harness engineering": 270,
"loop engineering": 62,
"prompt engineering": 1385
},
"2026-11": {
"context engineering": 63,
"harness engineering": 110,
"loop engineering": 42,
"prompt engineering": 463
}
},
"github_yearly_2023_2024": {
"context engineering": {
"2023": 0.0,
"2024": 10.0
},
"harness engineering": {
"2023": 3.0,
"2024": 3.0
},
"loop engineering": {
"2023": 1.0,
"2024": 0.0
},
"prompt engineering": {
"2023": 1624.0,
"2024": 2742.0
}
},
"github_term_summary": {
"context engineering": {
"sum_2025_01_to_2026_11": 2655,
"first_month_ge_10": "2025-06",
"peak_month": "2026-08",
"peak_n": 208
},
"harness engineering": {
"sum_2025_01_to_2026_11": 2469,
"first_month_ge_10": "2026-02",
"peak_month": "2026-04",
"peak_n": 408
},
"loop engineering": {
"sum_2025_01_to_2026_11": 1389,
"first_month_ge_10": "2026-05",
"peak_month": "2026-07",
"peak_n": 426
},
"prompt engineering": {
"sum_2025_01_to_2026_11": 19744,
"first_month_ge_10": "2025-01",
"peak_month": "2026-07",
"peak_n": 1898
}
},
"github_n_repos": 6278,
"github_communities": {
"context engineering": {
"repos": 2656,
"han_share": 0.035,
"org_share": 0.095,
"median_stars": 0.0,
"share_ge_100_stars": 0.023,
"top_languages": {
"Python": 940,
"TypeScript": 274,
"HTML": 161,
"JavaScript": 144,
"Jupyter Notebook": 143
},
"top_repos": [
{
"full_name": "dair-ai/Prompt-Engineering-Guide",
"stars": 80027,
"created": "2022-12-16"
},
{
"full_name": "gsd-build/get-shit-done",
"stars": 64144,
"created": "2025-12-14"
},
{
"full_name": "muratcankoylan/Agent-Skills-for-Context-Engineering",
"stars": 18345,
"created": "2025-12-21"
},
{
"full_name": "coleam00/context-engineering-intro",
"stars": 13914,
"created": "2025-07-02"
},
{
"full_name": "jasontang-ai/Context-Engineering",
"stars": 9281,
"created": "2025-06-29"
}
]
},
"harness engineering": {
"repos": 2279,
"han_share": 0.114,
"org_share": 0.069,
"median_stars": 0.0,
"share_ge_100_stars": 0.024,
"top_languages": {
"Python": 725,
"TypeScript": 276,
"Shell": 174,
"HTML": 170,
"JavaScript": 153
},
"top_repos": [
{
"full_name": "walkinglabs/learn-harness-engineering",
"stars": 24433,
"created": "2026-03-29"
},
{
"full_name": "liyupi/ai-guide",
"stars": 22070,
"created": "2025-02-13"
},
{
"full_name": "deusyu/harness-engineering",
"stars": 6284,
"created": "2026-03-23"
},
{
"full_name": "ModelEngine-Group/nexent",
"stars": 6192,
"created": "2025-04-28"
},
{
"full_name": "ai-boost/awesome-harness-engineering",
"stars": 5140,
"created": "2026-03-29"
}
]
},
"loop engineering": {
"repos": 1387,
"han_share": 0.087,
"org_share": 0.05,
"median_stars": 0.0,
"share_ge_100_stars": 0.012,
"top_languages": {
"Python": 493,
"JavaScript": 143,
"TypeScript": 135,
"HTML": 115,
"Shell": 111
},
"top_repos": [
{
"full_name": "HKUDS/DeepCode",
"stars": 16844,
"created": "2025-05-14"
},
{
"full_name": "cobusgreyling/loop-engineering",
"stars": 11763,
"created": "2026-06-09"
},
{
"full_name": "valkor-ai/loom",
"stars": 1488,
"created": "2026-06-09"
},
{
"full_name": "alchaincyf/loop-engineering-orange-book",
"stars": 1145,
"created": "2026-06-15"
},
{
"full_name": "dsta022/Loop-Engineering-for-VLA",
"stars": 845,
"created": "2026-05-30"
}
]
}
},
"github_cooccurrence": {
"context engineering": 2625,
"harness engineering": 2242,
"loop engineering": 1368,
"context engineering;harness engineering": 24,
"harness engineering;loop engineering": 12,
"context engineering;loop engineering": 6,
"context engineering;harness engineering;loop engineering": 1
},
"looparena": {
"type2": {
"no_control": 0.4917,
"fixed_control": 0.55,
"controller_mean": 0.5319,
"controller_best": "Claude Opus 4.8",
"controller_best_rate": 0.5875,
"gains_vs_no_control": {
"Fixed control": {
"gain_pp": 5.83,
"ci_pp": [
-3.08,
14.62
],
"ci_pp_conservative": [
-9.45,
20.74
],
"cost_multiple": 3.47
},
"Qwen3.6-35B-A3B": {
"gain_pp": 6.67,
"ci_pp": [
-2.25,
15.44
],
"ci_pp_conservative": [
-8.64,
21.54
],
"cost_multiple": 2.6
},
"Qwen3.7-Plus": {
"gain_pp": 7.5,
"ci_pp": [
-1.42,
16.25
],
"ci_pp_conservative": [
-7.81,
22.33
],
"cost_multiple": 3.73
},
"DeepSeek-V4-Flash-0731": {
"gain_pp": 1.67,
"ci_pp": [
-7.23,
10.52
],
"ci_pp_conservative": [
-13.52,
16.75
],
"cost_multiple": 6.1
},
"GLM 5.2": {
"gain_pp": -8.75,
"ci_pp": [
-17.45,
0.14
],
"ci_pp_conservative": [
-23.48,
6.54
],
"cost_multiple": 2.18
},
"GPT-5.5": {
"gain_pp": 7.5,
"ci_pp": [
-1.42,
16.25
],
"ci_pp_conservative": [
-7.81,
22.33
],
"cost_multiple": 8.52
},
"Claude Opus 4.8": {
"gain_pp": 9.58,
"ci_pp": [
0.67,
18.29
],
"ci_pp_conservative": [
-5.75,
24.3
],
"cost_multiple": 6.71
}
},
"no_control_cost_per_success": 3.3,
"cost_per_success": {
"No control": 3.3,
"Fixed control": 10.22,
"Qwen3.6-35B-A3B": 7.55,
"Qwen3.7-Plus": 10.65,
"DeepSeek-V4-Flash-0731": 19.45,
"GLM 5.2": 8.76,
"GPT-5.5": 24.35,
"Claude Opus 4.8": 18.51
}
},
"type3": {
"no_control": 0.275,
"fixed_control": 0.3125,
"controller_mean": 0.3174,
"controller_best": "GPT-5.5",
"controller_best_rate": 0.3625,
"gains_vs_no_control": {
"Fixed control": {
"gain_pp": 3.75,
"ci_pp": [
-4.39,
11.82
],
"ci_pp_conservative": [
-10.25,
17.56
],
"cost_multiple": 3.62
},
"Qwen3.6-35B-A3B": {
"gain_pp": 6.25,
"ci_pp": [
-1.99,
14.38
],
"ci_pp_conservative": [
-7.95,
20.12
],
"cost_multiple": 4.43
},
"Qwen3.7-Plus": {
"gain_pp": 6.67,
"ci_pp": [
-1.59,
14.8
],
"ci_pp_conservative": [
-7.57,
20.54
],
"cost_multiple": 4.85
},
"DeepSeek-V4-Flash-0731": {
"gain_pp": 0.83,
"ci_pp": [
-7.17,
8.82
],
"ci_pp_conservative": [
-12.92,
14.54
],
"cost_multiple": 12.49
},
"GLM 5.2": {
"gain_pp": -3.33,
"ci_pp": [
-11.11,
4.49
],
"ci_pp_conservative": [
-16.68,
10.17
],
"cost_multiple": 4.58
},
"GPT-5.5": {
"gain_pp": 8.75,
"ci_pp": [
0.41,
16.92
],
"ci_pp_conservative": [
-5.64,
22.65
],
"cost_multiple": 13.26
},
"Claude Opus 4.8": {
"gain_pp": 6.25,
"ci_pp": [
-1.99,
14.38
],
"ci_pp_conservative": [
-7.95,
20.12
],
"cost_multiple": 7.89
}
},
"no_control_cost_per_success": 9.51,
"cost_per_success": {
"No control": 9.51,
"Fixed control": 30.32,
"Qwen3.6-35B-A3B": 34.31,
"Qwen3.7-Plus": 37.13,
"DeepSeek-V4-Flash-0731": 115.25,
"GLM 5.2": 49.55,
"GPT-5.5": 95.67,
"Claude Opus 4.8": 61.12
}
},
"horizon_drop": {
"No control": {
"type2": 0.4917,
"type3": 0.275,
"relative_drop": 0.441,
"cost_ratio": 1.61
},
"Fixed control": {
"type2": 0.55,
"type3": 0.3125,
"relative_drop": 0.432,
"cost_ratio": 1.68
},
"Qwen3.6-35B-A3B": {
"type2": 0.5583,
"type3": 0.3375,
"relative_drop": 0.396,
"cost_ratio": 2.75
},
"Qwen3.7-Plus": {
"type2": 0.5667,
"type3": 0.3417,
"relative_drop": 0.397,
"cost_ratio": 2.1
},
"DeepSeek-V4-Flash-0731": {
"type2": 0.5083,
"type3": 0.2833,
"relative_drop": 0.443,
"cost_ratio": 3.3
},
"GLM 5.2": {
"type2": 0.4042,
"type3": 0.2417,
"relative_drop": 0.402,
"cost_ratio": 3.38
},
"GPT-5.5": {
"type2": 0.5667,
"type3": 0.3625,
"relative_drop": 0.36,
"cost_ratio": 2.51
},
"Claude Opus 4.8": {
"type2": 0.5875,
"type3": 0.3375,
"relative_drop": 0.426,
"cost_ratio": 1.9
}
},
"typeI_vs_typeIII_spearman": {
"rho": 0.696,
"p": 0.125,
"n": 6
},
"relative_gain_best": {
"type2": 0.195,
"type3": 0.318
}
},
"model_e1": {
"moderate": {
"n1": 0.36,
"n20": 0.8,
"ceiling": 0.8,
"n20_heterogeneous": 0.682
},
"strong": {
"n1": 0.38,
"n20": 0.969,
"ceiling": 0.969,
"n20_heterogeneous": 0.861
},
"weak": {
"n1": 0.32,
"n20": 0.571,
"ceiling": 0.571,
"n20_heterogeneous": 0.512
}
},
"model_e2_T300": {
"append_kappa0.1": 5.95,
"append_kappa1.0": 52.72,
"mask_kappa0.1": 3.17,
"mask_kappa1.0": 16.21,
"compact_kappa0.1": 2.88,
"compact_kappa1.0": 20.66,
"reset_kappa0.1": 2.72,
"reset_kappa1.0": 16.57
},
"model_e2_T100": {
"append_kappa0.1": 1.5,
"append_kappa1.0": 12.72,
"mask_kappa0.1": 0.92,
"mask_kappa1.0": 4.64,
"compact_kappa0.1": 0.91,
"compact_kappa1.0": 6.36,
"reset_kappa0.1": 0.69,
"reset_kappa1.0": 4.03
},
"model_e3_switch_T_by_gamma": {
"0.0": 150,
"0.5": 100,
"1.0": 60,
"2.0": 40,
"3.0": 40,
"4.0": 40
},
"model_e3_gamma2": {
"append": {
"10": 0.89,
"20": 0.782,
"40": 0.568,
"60": 0.402,
"80": 0.283,
"100": 0.192,
"150": 0.056,
"200": 0.014,
"300": 0.0
},
"compact": {
"10": 0.89,
"20": 0.782,
"40": 0.568,
"60": 0.406,
"80": 0.286,
"100": 0.2,
"150": 0.076,
"200": 0.024,
"300": 0.002
},
"mask": {
"10": 0.89,
"20": 0.782,
"40": 0.558,
"60": 0.38,
"80": 0.282,
"100": 0.193,
"150": 0.095,
"200": 0.042,
"300": 0.015
},
"reset": {
"10": 0.89,
"20": 0.782,
"40": 0.594,
"60": 0.464,
"80": 0.35,
"100": 0.28,
"150": 0.168,
"200": 0.108,
"300": 0.045
}
},
"model_e4": {
"independent|0.0": {
"append": 0.079,
"compact": 0.072,
"mask": 0.069,
"reset": 0.124
},
"independent|0.25": {
"append": 0.118,
"compact": 0.104,
"mask": 0.105,
"reset": 0.176
},
"independent|0.5": {
"append": 0.191,
"compact": 0.201,
"mask": 0.213,
"reset": 0.279
},
"independent|0.75": {
"append": 0.368,
"compact": 0.374,
"mask": 0.411,
"reset": 0.483
},
"independent|0.9": {
"append": 0.626,
"compact": 0.655,
"mask": 0.657,
"reset": 0.725
},
"self-critique|0.0": {
"append": 0.079,
"compact": 0.072,
"mask": 0.069,
"reset": 0.124
},
"self-critique|0.25": {
"append": 0.12,
"compact": 0.09,
"mask": 0.09,
"reset": 0.166
},
"self-critique|0.5": {
"append": 0.187,
"compact": 0.153,
"mask": 0.135,
"reset": 0.252
},
"self-critique|0.75": {
"append": 0.336,
"compact": 0.242,
"mask": 0.213,
"reset": 0.396
},
"self-critique|0.9": {
"append": 0.533,
"compact": 0.351,
"mask": 0.32,
"reset": 0.557
}
},
"model_e4_logodds_gap_reset_minus_append": {
"0.0": 0.496,
"0.25": 0.467,
"0.5": 0.496,
"0.75": 0.471,
"0.9": 0.454
},
"model_e4_selfcritique_penalty_d09": {
"append": 0.093,
"compact": 0.304,
"mask": 0.337,
"reset": 0.167
},
"model_e5": {
"eps0.005|gamma0.5|d0.5": {
"reset_minus_append": -0.038,
"best": "append"
},
"eps0.005|gamma0.5|d0.9": {
"reset_minus_append": -0.022,
"best": "append"
},
"eps0.005|gamma2.0|d0.5": {
"reset_minus_append": 0.019,
"best": "reset"
},
"eps0.005|gamma2.0|d0.9": {
"reset_minus_append": 0.007,
"best": "reset"
},
"eps0.01|gamma0.5|d0.5": {
"reset_minus_append": -0.022,
"best": "append"
},
"eps0.01|gamma0.5|d0.9": {
"reset_minus_append": -0.008,
"best": "append"
},
"eps0.01|gamma2.0|d0.5": {
"reset_minus_append": 0.073,
"best": "reset"
},
"eps0.01|gamma2.0|d0.9": {
"reset_minus_append": 0.038,
"best": "reset"
},
"eps0.02|gamma0.5|d0.5": {
"reset_minus_append": 0.0,
"best": "append"
},
"eps0.02|gamma0.5|d0.9": {
"reset_minus_append": 0.011,
"best": "reset"
},
"eps0.02|gamma2.0|d0.5": {
"reset_minus_append": 0.083,
"best": "reset"
},
"eps0.02|gamma2.0|d0.9": {
"reset_minus_append": 0.095,
"best": "reset"
},
"eps0.04|gamma0.5|d0.5": {
"reset_minus_append": 0.027,
"best": "reset"
},
"eps0.04|gamma0.5|d0.9": {
"reset_minus_append": 0.033,
"best": "reset"
},
"eps0.04|gamma2.0|d0.5": {
"reset_minus_append": 0.06,
"best": "reset"
},
"eps0.04|gamma2.0|d0.9": {
"reset_minus_append": 0.149,
"best": "reset"
}
},
"model_e5_verifier_gain_append": {
"eps0.005|gamma0.5": 0.215,
"eps0.005|gamma2.0": 0.265,
"eps0.01|gamma0.5": 0.308,
"eps0.01|gamma2.0": 0.377,
"eps0.02|gamma0.5": 0.423,
"eps0.02|gamma2.0": 0.434,
"eps0.04|gamma0.5": 0.456,
"eps0.04|gamma2.0": 0.344
},
"model_e6_best_success_share": {
"30": {
"append": 0.707,
"compact": 0.103,
"mask": 0.05,
"reset": 0.14
},
"100": {
"append": 0.207,
"compact": 0.15,
"mask": 0.247,
"reset": 0.397
}
},
"model_e6_best_cost_share": {
"30": {
"append": 0.443,
"compact": 0.143,
"mask": 0.0,
"reset": 0.413
},
"100": {
"append": 0.02,
"compact": 0.168,
"mask": 0.077,
"reset": 0.734
}
},
"model_e6_retention_grid_best": {
"0.02": {
"0.02": "reset",
"0.05": "reset",
"0.1": "reset",
"0.2": "reset",
"0.4": "reset"
},
"0.05": {
"0.02": "reset",
"0.05": "reset",
"0.1": "reset",
"0.2": "reset",
"0.4": "reset"
},
"0.1": {
"0.02": "reset",
"0.05": "reset",
"0.1": "reset",
"0.2": "reset",
"0.4": "reset"
},
"0.2": {
"0.02": "compact",
"0.05": "compact",
"0.1": "compact",
"0.2": "reset",
"0.4": "reset"
},
"0.4": {
"0.02": "compact",
"0.05": "compact",
"0.1": "compact",
"0.2": "mask",
"0.4": "mask"
}
},
"model_e6_tree_T100": {
"text": "|--- lam_r <= 0.17\n| |--- s <= 1764.83\n| | |--- class: append\n| |--- s > 1764.83\n| | |--- delta_mask <= 0.13\n| | | |--- class: mask\n| | |--- delta_mask > 0.13\n| | | |--- class: reset\n|--- lam_r > 0.17\n| |--- delta_mask <= 0.27\n| | |--- s <= 2387.65\n| | | |--- class: append\n| | |--- s > 2387.65\n| | | |--- class: mask\n| |--- delta_mask > 0.27\n| | |--- lam_c <= 0.11\n| | | |--- class: compact\n| | |--- lam_c > 0.11\n| | | |--- class: reset\n",
"train_accuracy": 0.64,
"importances": {
"gamma": 0.0,
"eps_mean": 0.0,
"rho": 0.0,
"delta_mask": 0.287,
"lam_c": 0.19,
"lam_r": 0.297,
"d": 0.0,
"k": 0.0,
"H": 0.0,
"s": 0.226
}
},
"model_e6_lowest_loss_rule_accuracy": 0.655,
"model_e6_lowest_loss_rule_n": 238,
"audit": {
"round1_v1_layer_kappa": 0.663,
"round1_v1_contribution_kappa": 0.781,
"round2_v2_layer_kappa": 0.856,
"round2_v2_layer_agreement": 0.9,
"round2_v2_contribution_kappa": 0.764,
"pooled_v2_layer_kappa": 0.859,
"n_round1": 100,
"n_round2": 60
}
}
{
"flow": {
"lists": 5,
"entries": 1891,
"unique": 1492,
"excluded": 97,
"excluded_by_reason": {
"excluded section": 58,
"out of scope": 31,
"model architecture": 8
},
"included": 1395,
"entries_by_list": {
"aiboost_harness": 532,
"deepjlu_graph": 447,
"picrew_harness": 385,
"meirtz_context": 348,
"walkinglabs_harness": 179
},
"multi_list_sources": 180
},
"composition": {
"academic": {
"context": 190,
"evaluation": 92,
"harness": 110,
"loop": 152
},
"grey": {
"context": 39,
"evaluation": 38,
"harness": 146,
"loop": 51
},
"repository": {
"context": 61,
"evaluation": 70,
"harness": 346,
"loop": 100
}
},
"composition_layer_totals": {
"harness": 602,
"loop": 303,
"context": 290,
"evaluation": 200
},
"composition_type_totals": {
"repository": 577,
"academic": 544,
"grey": 274
},
"academic_dated_share": 0.945,
"timeline_academic": {
"context": {
"2020H1": 3,
"2020H2": 2,
"2021H1": 1,
"2021H2": 4,
"2022H1": 2,
"2022H2": 10,
"2023H1": 16,
"2023H2": 28,
"2024H1": 40,
"2024H2": 27,
"2025H1": 23,
"2025H2": 4,
"2026H1": 25,
"2026H2": 4
},
"evaluation": {
"2020H1": 0,
"2020H2": 0,
"2021H1": 0,
"2021H2": 0,
"2022H1": 0,
"2022H2": 0,
"2023H1": 2,
"2023H2": 8,
"2024H1": 14,
"2024H2": 6,
"2025H1": 4,
"2025H2": 4,
"2026H1": 23,
"2026H2": 10
},
"harness": {
"2020H1": 1,
"2020H2": 0,
"2021H1": 0,
"2021H2": 0,
"2022H1": 3,
"2022H2": 1,
"2023H1": 6,
"2023H2": 6,
"2024H1": 12,
"2024H2": 9,
"2025H1": 15,
"2025H2": 6,
"2026H1": 34,