devision / evaluation /results.json
lukbit's picture
deVision v0.2 (github model-v0.2, 5437ac96b022e0c7029c40100575733dc852a707)
d8530d4 verified
Raw History Blame Contribute Delete
2.82 kB
{
"temperature": {
"choice": 1.9567745923995972,
"noul": 1.6382114887237549,
"by_options": {
"choice:2": 1.8836495876312256,
"choice:3-5": 1.979034662246704,
"noul:2": 1.6382114887237549
}
},
"test_sets": {
"test_exist": {
"questions": 1120,
"accuracy": 0.9384,
"nll": 0.162,
"ece": 0.0204,
"accuracy_mismatched": 0.4929
},
"test_vqa_choice": {
"questions": 1420,
"accuracy": 0.8979,
"nll": 0.2706,
"ece": 0.0262,
"accuracy_mismatched": 0.3993
},
"test_size": {
"questions": 1100,
"accuracy": 0.86,
"nll": 0.3503,
"ece": 0.036,
"accuracy_mismatched": 0.5045
},
"bench_pope": {
"questions": 8676,
"accuracy": 0.8511,
"nll": 0.3702,
"ece": 0.0578,
"accuracy_mismatched": 0.527
},
"test_gqa": {
"questions": 992,
"accuracy": 0.7843,
"nll": 0.409,
"ece": 0.0419,
"accuracy_mismatched": 0.5323
},
"test_position": {
"questions": 1274,
"accuracy": 0.8721,
"nll": 0.3045,
"ece": 0.0251,
"accuracy_mismatched": 0.4929
},
"test_vqa_yesno": {
"questions": 1000,
"accuracy": 0.71,
"nll": 0.5768,
"ece": 0.0223,
"accuracy_mismatched": 0.518
},
"test_relation": {
"questions": 1950,
"accuracy": 0.7108,
"nll": 0.5401,
"ece": 0.0471,
"accuracy_mismatched": 0.4841
},
"test_vsr": {
"questions": 904,
"accuracy": 0.6792,
"nll": 0.6155,
"ece": 0.0478,
"accuracy_mismatched": 0.4812
},
"test_v7w": {
"questions": 1000,
"accuracy": 0.721,
"nll": 0.6838,
"ece": 0.0302,
"accuracy_mismatched": 0.422
},
"test_count_fresh": {
"questions": 600,
"accuracy": 0.74,
"nll": 0.5763,
"ece": 0.0648,
"accuracy_mismatched": 0.5067
}
},
"laya_vision": {
"vqav2_yesno": {
"questions": 4887,
"laya_vision": 0.7172,
"devision": 0.7248,
"difference": 0.0076,
"ci95": [
-0.0087,
0.0233
]
},
"aokvqa": {
"questions": 1138,
"laya_vision": 0.5984,
"devision": 0.6257,
"difference": 0.0272,
"ci95": [
-0.0062,
0.0598
]
},
"scienceqa": {
"questions": 2097,
"laya_vision": 0.824,
"devision": 0.7663,
"difference": -0.0577,
"ci95": [
-0.0792,
-0.0362
]
},
"scienceqa_natural_needs_picture": {
"questions": 323,
"laya_vision": 0.6997,
"devision": 0.4551,
"difference": -0.2446,
"ci95": [
-0.3158,
-0.1765
]
},
"pope": {
"random": {
"laya_vision_published": 0.836,
"devision": 0.8913
},
"popular": {
"laya_vision_published": 0.819,
"devision": 0.868
},
"adversarial": {
"laya_vision_published": 0.777,
"devision": 0.7913
}
}
},
"cpu_latency": {
"p50": 148,
"p95": 161,
"threads": 4,
"machine": "Darwin arm64",
"source": "runs/v12-ext/cpu/single.json"
}
}