kwvtSA9ed3 / data /crossconfig_comparison.json
DineshAI's picture
All six claims decided with reproducible evidence (5 VERIFIED, 1 FALSIFIED as stated in the main text)
e2d54c9 verified
Raw
History Blame Contribute Delete
1.17 kB
{
"detail": [
{
"best_pluralistic": "balanced pluralistic (M=5)",
"best_single": "single-reward (M=1, target class 0)",
"margin": 1.1427,
"metric": "entropy_tail_mean",
"min_margin_required": 0.05,
"pluralistic": {
"balanced pluralistic (M=2)": 0.0,
"balanced pluralistic (M=5)": 1.1427
},
"single_reward": {
"single-reward (M=1, target class 0)": 0.0
}
},
{
"best_pluralistic": "two preferences (T_A=10, T_B=15, d=5)",
"best_single": "single preference (T_A=10)",
"margin": 2.4241,
"metric": "H_tail_mean",
"min_margin_required": 0.05,
"pluralistic": {
"two preferences (T_A=10, T_B=15, d=5)": 3.1001
},
"single_reward": {
"single preference (T_A=10)": 0.676
}
}
],
"source_runs": [
"3e518ee8-236a-46c0-8222-94ede16fd6bf",
"1b461517-66f6-4b9a-8b61-2557b2e46270",
"8abf3220-3efa-4a6a-a3e3-5150e4465e27",
"79d9795a-1f93-4544-9281-b5c0c4ba7403",
"19b9ba40-a215-4d32-a158-e0fefce956e6"
],
"verdicts": [
"claim5/pluralistic-vs-single",
"claim6/pluralistic-vs-single"
]
}