devision / evaluation /results.json
lukbit's picture
deVision v0.5 (github model-v0.5, 70adc4a7734ae63f9e1080733049b1cdd9b830c4)
3f36a55 verified
Raw History Blame Contribute Delete
3.25 kB
{
"temperature": {
"choice": 1.160062551498413,
"noul": 0.9890053272247314,
"by_options": {
"choice:2": 1.1245635747909546,
"choice:3-5": 1.1677029132843018,
"noul:2": 0.9890053272247314
}
},
"test_sets": {
"test_exist": {
"questions": 1120,
"accuracy": 0.9518,
"nll": 0.1888,
"ece": 0.0705,
"accuracy_mismatched": 0.5054
},
"test_vqa_choice": {
"questions": 1453,
"accuracy": 0.905,
"nll": 0.2511,
"ece": 0.0265,
"accuracy_mismatched": 0.373
},
"test_size": {
"questions": 1198,
"accuracy": 0.8765,
"nll": 0.4159,
"ece": 0.1535,
"accuracy_mismatched": 0.5109
},
"bench_pope": {
"questions": 9000,
"accuracy": 0.8607,
"nll": 0.3615,
"ece": 0.0285,
"accuracy_mismatched": 0.5177
},
"test_gqa": {
"questions": 986,
"accuracy": 0.8448,
"nll": 0.3462,
"ece": 0.0291,
"accuracy_mismatched": 0.5578
},
"test_position": {
"questions": 1276,
"accuracy": 0.9412,
"nll": 0.1818,
"ece": 0.0407,
"accuracy_mismatched": 0.5
},
"test_vqa_yesno": {
"questions": 1000,
"accuracy": 0.721,
"nll": 0.5673,
"ece": 0.0622,
"accuracy_mismatched": 0.501
},
"test_relation": {
"questions": 2366,
"accuracy": 0.9121,
"nll": 0.2263,
"ece": 0.0281,
"accuracy_mismatched": 0.4721
},
"test_vsr": {
"questions": 902,
"accuracy": 0.7184,
"nll": 0.5901,
"ece": 0.1022,
"accuracy_mismatched": 0.469
},
"test_v7w": {
"questions": 1000,
"accuracy": 0.777,
"nll": 0.5846,
"ece": 0.0407,
"accuracy_mismatched": 0.446
},
"bench_countbench": {
"questions": 491,
"accuracy": 0.5153,
"nll": 1.2606,
"ece": 0.0791,
"accuracy_mismatched": 0.1059
},
"test_position_flip": {
"questions": 640,
"accuracy": 0.9406,
"nll": 0.1774,
"ece": 0.0337,
"accuracy_mismatched": 0.5219
},
"test_relation_flip": {
"questions": 1786,
"accuracy": 0.9227,
"nll": 0.2098,
"ece": 0.0196,
"accuracy_mismatched": 0.4905
},
"bench_whatsup": {
"questions": 820,
"accuracy": 0.5183,
"nll": 1.1743,
"ece": 0.096,
"accuracy_mismatched": 0.2329
}
},
"laya_vision": {
"vqav2_yesno": {
"questions": 4887,
"laya_vision": 0.7172,
"devision": 0.7493,
"difference": 0.0321,
"ci95": [
0.0174,
0.0469
]
},
"aokvqa": {
"questions": 1138,
"laya_vision": 0.5984,
"devision": 0.6274,
"difference": 0.029,
"ci95": [
-0.0026,
0.0624
]
},
"scienceqa": {
"questions": 2097,
"laya_vision": 0.824,
"devision": 0.8207,
"difference": -0.0033,
"ci95": [
-0.0234,
0.0153
]
},
"scienceqa_natural_needs_picture": {
"questions": 323,
"laya_vision": 0.6997,
"devision": 0.6006,
"difference": -0.0991,
"ci95": [
-0.1641,
-0.0341
]
},
"pope": {
"random": {
"laya_vision_published": 0.836,
"devision": 0.8943
},
"popular": {
"laya_vision_published": 0.819,
"devision": 0.867
},
"adversarial": {
"laya_vision_published": 0.777,
"devision": 0.8207
}
}
},
"cpu_latency": {
"p50": 301,
"p95": 321,
"threads": 4,
"machine": "Linux x86_64",
"source": "release/v0.5/cpu_latency.json"
}
}