v3.9.0 发版:规则库补强(E3 +20 条,548→568)+ 收敛层防护定型
This commit is contained in:
@@ -1,13 +1,13 @@
|
||||
{
|
||||
"model": "kimi-k3",
|
||||
"skill_mode": "with-skill",
|
||||
"skill_version": "3.6.2-未发版(规则库 3.7)",
|
||||
"skill_version": "3.9.0-未发版(规则库 568 条)",
|
||||
"eval_date": "2026-10-13",
|
||||
"benchmark": "BaziQA-全球算命师大赛",
|
||||
"overall": {
|
||||
"total": 200,
|
||||
"correct": 79,
|
||||
"accuracy": 39.5,
|
||||
"correct": 81,
|
||||
"accuracy": 40.5,
|
||||
"no_answer": 0
|
||||
},
|
||||
"baseline": {
|
||||
@@ -51,9 +51,9 @@
|
||||
},
|
||||
"contest8_2023": {
|
||||
"total": 40,
|
||||
"correct": 12,
|
||||
"correct": 17,
|
||||
"no_answer": 0,
|
||||
"accuracy": 30.0
|
||||
"accuracy": 42.5
|
||||
},
|
||||
"contest8_2024": {
|
||||
"total": 40,
|
||||
@@ -63,47 +63,47 @@
|
||||
},
|
||||
"contest8_2025": {
|
||||
"total": 40,
|
||||
"correct": 17,
|
||||
"correct": 14,
|
||||
"no_answer": 0,
|
||||
"accuracy": 42.5
|
||||
"accuracy": 35.0
|
||||
}
|
||||
},
|
||||
"per_dimension": {
|
||||
"流年运势": {
|
||||
"total": 37,
|
||||
"correct": 13,
|
||||
"correct": 16,
|
||||
"no_answer": 0,
|
||||
"accuracy": 35.1
|
||||
"accuracy": 43.2
|
||||
},
|
||||
"感情": {
|
||||
"total": 49,
|
||||
"correct": 19,
|
||||
"correct": 17,
|
||||
"no_answer": 0,
|
||||
"accuracy": 38.8
|
||||
"accuracy": 34.7
|
||||
},
|
||||
"六亲": {
|
||||
"total": 29,
|
||||
"correct": 13,
|
||||
"correct": 12,
|
||||
"no_answer": 0,
|
||||
"accuracy": 44.8
|
||||
"accuracy": 41.4
|
||||
},
|
||||
"事业": {
|
||||
"total": 25,
|
||||
"correct": 10,
|
||||
"correct": 12,
|
||||
"no_answer": 0,
|
||||
"accuracy": 40.0
|
||||
"accuracy": 48.0
|
||||
},
|
||||
"性格": {
|
||||
"total": 20,
|
||||
"correct": 10,
|
||||
"correct": 12,
|
||||
"no_answer": 0,
|
||||
"accuracy": 50.0
|
||||
"accuracy": 60.0
|
||||
},
|
||||
"学业": {
|
||||
"total": 12,
|
||||
"correct": 6,
|
||||
"correct": 5,
|
||||
"no_answer": 0,
|
||||
"accuracy": 50.0
|
||||
"accuracy": 41.7
|
||||
},
|
||||
"财富": {
|
||||
"total": 13,
|
||||
@@ -113,9 +113,9 @@
|
||||
},
|
||||
"健康": {
|
||||
"total": 13,
|
||||
"correct": 5,
|
||||
"correct": 4,
|
||||
"no_answer": 0,
|
||||
"accuracy": 38.5
|
||||
"accuracy": 30.8
|
||||
},
|
||||
"其他": {
|
||||
"total": 2,
|
||||
@@ -124,5 +124,5 @@
|
||||
"accuracy": 0.0
|
||||
}
|
||||
},
|
||||
"notes": "判分:评测/工具/run_eval.py;求解留档:评测/ai答案/kimi-k3/with-skill/(41 位命主,规则库 3.7 重跑,2026-10-13 收录;3.6 版留档存档于 with-skill-v3.6/);官方基线口径见 试题/BaziQA-全球算命师大赛/SOURCE.md"
|
||||
"notes": "判分:评测/工具/run_eval.py;求解留档:评测/ai答案/kimi-k3/with-skill/(41 位命主:规则库 3.7 全量 + 3.9 补强重跑 17 位命主的合并集,2026-10-13 收录;3.6 版留档存档于 with-skill-v3.6/);官方基线口径见 试题/BaziQA-全球算命师大赛/SOURCE.md"
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user