canary-contracts / eval_results_dev.json
NagaYu's picture
canary-contracts: engine (canary.py, sha256 8319b050), contract packs, quickstart, evaluation reports
01ef499 verified
Raw History Blame Contribute Delete
1.16 kB
{
"cases": 306,
"accuracy": 1.0,
"fail_recall": 1.0,
"false_alarm_rate": 0.0,
"per_split": {
"dev": {
"n": 306,
"accuracy": 1.0
}
},
"per_contract_type": {
"grounded": {
"n": 24,
"accuracy": 1.0
},
"json_valid": {
"n": 22,
"accuracy": 1.0
},
"max_tokens": {
"n": 1,
"accuracy": 1.0
},
"must_contain": {
"n": 7,
"accuracy": 1.0
},
"must_not_contain": {
"n": 7,
"accuracy": 1.0
},
"must_refuse": {
"n": 70,
"accuracy": 1.0
},
"no_pii": {
"n": 89,
"accuracy": 1.0
},
"no_secrets": {
"n": 59,
"accuracy": 1.0
},
"no_system_leak": {
"n": 21,
"accuracy": 1.0
},
"schema": {
"n": 4,
"accuracy": 1.0
},
"stable": {
"n": 2,
"accuracy": 1.0
}
},
"per_lang": {
"en": {
"n": 255,
"accuracy": 1.0
},
"ja": {
"n": 51,
"accuracy": 1.0
}
},
"confusion": {
"fail->fail": 157,
"pass->pass": 148,
"skipped->skipped": 1
},
"misses": [],
"engine": "Canary 1.0.0"
}