Adam1010's picture
v1.1: Document financial domain findings - strongest Goodhart Gap detector
dc1a656 verified
{
"total_problems": 101,
"domains": {
"math_discount": 15,
"time": 13,
"recipe": 7,
"financial": 10,
"units": 7,
"scheduling": 7,
"logic": 8,
"spatial": 7,
"procedural": 6,
"text": 7,
"sequence": 7,
"causal": 7
},
"difficulty_distribution": {
"easy": 46,
"medium": 48,
"hard": 7
},
"step_distribution": {
"2": 69,
"3": 22,
"4": 7,
"6": 1,
"1": 2
}
}