sackfield's picture
Inputs that papers and analyses read (#3255) (part 4)
7e85ee4 verified
Raw History Blame Contribute Delete
543 Bytes
{
"outcome": "NEEDS_REVISION",
"probe_verified": false,
"gates": {
"enough_hard_negatives": true,
"enough_holdouts": true,
"specificity": true,
"sensitivity": false,
"counterfactual_reproduction": true,
"lookahead_consistent": true,
"independent_review": true,
"beats_confound_baseline": false,
"not_default_on": true
},
"specificity": 0.916667,
"sensitivity": 0.428571,
"balanced_accuracy": 0.672619,
"active_probes": 4,
"measured_probes": 20,
"reviewer": "external/claude-sonnet-5-5"
}