refract-hsmall-blend2 / eval /eval_sft_v5.json
Nikhil0097's picture
Initial upload: weights + model card
263e9a6 verified
Raw
History Blame Contribute Delete
346 Bytes
{
"label": "sft",
"n_scored": 20,
"n_fail": 0,
"judge_means": {
"admits_multiplicity": 4.5,
"distinctness": 3.4,
"concreteness": 3.25,
"decisiveness": 3.95,
"foresight": 2.8,
"viability": 2.2
},
"judge_overall_mean": 3.35,
"viability_pass(>=3)": 30.0,
"distinct_pass(>=4)": 50.0,
"full_pd_diag": 0.3206
}