captionbert-8192-v2-b / deep-arm /deep_arm_bench.json
AbstractPhil's picture
deep-arm: certified 8-task bench (engaged mode)
a156c4b verified
Raw
History Blame Contribute Delete
1.31 kB
{
"rows": {
"captionbert-b + deep-arm": {
"STS-B": {
"spearman": 0.5428209488317594,
"self_cos": 0.288449227809906,
"erank": 37.337919076587866
},
"SICK-R": {
"spearman": 0.6263769719404698,
"self_cos": 0.38013410568237305,
"erank": 40.155722513594505
},
"STS12": {
"spearman": 0.4714098311002296,
"self_cos": 0.3429536521434784,
"erank": 24.587241917590244
},
"STS13": {
"spearman": 0.5747958345812133,
"self_cos": 0.3156481683254242,
"erank": 27.401931698686575
},
"STS14": {
"spearman": 0.5231269302465135,
"self_cos": 0.28109976649284363,
"erank": 39.43578755984569
},
"STS15": {
"spearman": 0.6792496059354218,
"self_cos": 0.3246782124042511,
"erank": 26.26469335307641
},
"STS16": {
"spearman": 0.6190642731967696,
"self_cos": 0.3443126678466797,
"erank": 27.436889390044094
},
"BIOSSES": {
"spearman": 0.487143630018559,
"self_cos": 0.7020799517631531,
"erank": 27.60395066924268
}
}
},
"mean": 0.5654985032313671,
"delta_vs_bare": -0.037594436017591715
}