kodama-vision / evaluation /shared_image_comparison.json
Cem13's picture
Expand usage, use cases, performance details, and runnable examples
9524435 verified
Raw History Blame Contribute Delete
3.51 kB
{
"newsfin_short": {
"omni": {
"n": 140,
"accuracy": 0.6642857142857143,
"nll": 1.0239277480868136,
"ece": 0.086806222512248,
"median_ms": 152.98170200549066
},
"existing": {
"n": 140,
"accuracy": 0.6357142857142857,
"nll": 0.8810642035031887,
"ece": 0.09626927165913261,
"median_ms": 31.578160007484257
}
},
"newsfin_long": {
"omni": {
"n": 56,
"accuracy": 0.5535714285714286,
"nll": 1.0747286532578382,
"ece": 0.20183825996408222,
"median_ms": 399.9315219989512
},
"existing": {
"n": 56,
"accuracy": 0.7142857142857143,
"nll": 0.7747561391758683,
"ece": 0.11390295787714419,
"median_ms": 76.97567199647892
}
},
"objects": {
"omni": {
"n": 48,
"accuracy": 0.8958333333333334,
"nll": 0.3463743257572875,
"ece": 0.1607140380942711,
"median_ms": 181.64109149074648
},
"existing": {
"n": 48,
"accuracy": 0.875,
"nll": 0.4381676075638272,
"ece": 0.06802068458637223,
"median_ms": 87.48945897968952
}
},
"cars": {
"omni": {
"n": 48,
"accuracy": 0.8333333333333334,
"nll": 0.8116724576537026,
"ece": 0.1538433370367329,
"median_ms": 185.9130049997475
},
"existing": {
"n": 48,
"accuracy": 0.9375,
"nll": 0.13429676287610173,
"ece": 0.06006058381414428,
"median_ms": 72.42575800046325
}
},
"documents": {
"omni": {
"n": 64,
"accuracy": 0.78125,
"nll": 0.5672011287347207,
"ece": 0.1941846296614581,
"median_ms": 177.84590501105413
},
"existing": {
"n": 64,
"accuracy": 0.578125,
"nll": 1.3456605204542036,
"ece": 0.19807713711634278,
"median_ms": 46.53450699697714
}
},
"controlled_charts": {
"omni": {
"n": 8,
"accuracy": 1.0,
"nll": 0.03572675939492173,
"ece": 0.03509586709753065,
"median_ms": 278.0658479896374
},
"existing": {
"n": 8,
"accuracy": 0.0,
"nll": 2.4772741884888907,
"ece": 0.7510761469602585,
"median_ms": 48.076307502924465
}
},
"controlled_pdf": {
"omni": {
"n": 8,
"accuracy": 1.0,
"nll": 0.039342390823764464,
"ece": 0.038576512512841776,
"median_ms": 293.2979239994893
},
"existing": {
"n": 8,
"accuracy": 0.0,
"nll": 2.510732355757172,
"ece": 0.7607670575380325,
"median_ms": 49.164392505190335
}
},
"controlled_speech": {
"omni": {
"n": 8,
"accuracy": 1.0,
"nll": 0.07976032458653613,
"ece": 0.0751707394894201,
"median_ms": 320.00854500802234
},
"existing": {
"n": 0
}
},
"controlled_video_order": {
"omni": {
"n": 8,
"accuracy": 0.625,
"nll": 0.662364327099227,
"ece": 0.2687584065892079,
"median_ms": 563.8657774979947
},
"existing": {
"n": 0
}
},
"controlled_video_audio": {
"omni": {
"n": 8,
"accuracy": 1.0,
"nll": 0.08512731488755165,
"ece": 0.08009641117628796,
"median_ms": 572.0832340011839
},
"existing": {
"n": 0
}
},
"environmental_audio": {
"omni": {
"n": 20,
"accuracy": 1.0,
"nll": 0.06514317194261468,
"ece": 0.05566597857985209,
"median_ms": 348.48555049393326
},
"existing": {
"n": 0
}
}
}