3266 lines
75 KiB
JSON
3266 lines
75 KiB
JSON
{
|
|
"timestamp": "2026-03-05T18:14:19.801264+00:00",
|
|
"git_sha": "6c7598f43",
|
|
"doc_count": 171,
|
|
"pipeline_count": 1,
|
|
"aggregates": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"mean_sf1": 0.1946004540594232,
|
|
"mean_tf1": 0.7919178393306505,
|
|
"mean_time_ms": 5420.186888643275,
|
|
"p50_sf1": 0.09610431700415677,
|
|
"p50_tf1": 0.9476351351351352,
|
|
"p50_time_ms": 91.420083,
|
|
"p90_time_ms": 3635.487541
|
|
}
|
|
],
|
|
"docs": [
|
|
{
|
|
"name": "2023-06-20-PV",
|
|
"file_size": 35535,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.06834381476167703,
|
|
"tf1": 0.9730586370839936,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Image": 0.0,
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.8201257771401244,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 230.53725
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "2203.01017v2",
|
|
"file_size": 7215177,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.07900260437326329,
|
|
"tf1": 0.9918046685129712,
|
|
"order_score": 0.90625,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Paragraph": 0.4903487205943379,
|
|
"ListItem": 0.42634777687556896,
|
|
"H1": 0.0,
|
|
"Table": 0.0,
|
|
"Code": 0.0
|
|
},
|
|
"time_ms": 1052.6
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "2206.01062",
|
|
"file_size": 4310680,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.01888807877702598,
|
|
"tf1": 0.8990441272751081,
|
|
"order_score": 0.7741935483870968,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.2716093496103993,
|
|
"H1": 0.0,
|
|
"ListItem": 0.03418803418803419,
|
|
"H2": 0.0,
|
|
"H3": 0.0,
|
|
"Code": 0.0
|
|
},
|
|
"time_ms": 1158.3007499999999
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "2305.03393v1",
|
|
"file_size": 4318934,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.41037881842434326,
|
|
"tf1": 0.9751671442215856,
|
|
"order_score": 0.9814814814814815,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"Paragraph": 0.686086438951411,
|
|
"H2": 1.0,
|
|
"Image": 0.0,
|
|
"ListItem": 0.8888888888888891,
|
|
"H1": 0.33333333333333337,
|
|
"H3": 0.0
|
|
},
|
|
"time_ms": 298.077791
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "2305.03393v1-pg9",
|
|
"file_size": 162131,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.050681026290782395,
|
|
"tf1": 0.9973262032085561,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Table": 0.0,
|
|
"Paragraph": 0.7095343680709535,
|
|
"H1": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 38.361582999999996
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "4833695495684096",
|
|
"file_size": 3759,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.14545454545454545,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Paragraph": 0.7272727272727273
|
|
},
|
|
"time_ms": 0.700833
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "RECEIPT-2024-TXN-98765_retail_purchase",
|
|
"file_size": 8459,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.04477992485238861,
|
|
"tf1": 0.9716088328075709,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.2686795491143317,
|
|
"Image": 0.0,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 15.063583
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "REPAIR-2022-INV-001_multipage",
|
|
"file_size": 176629,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.3356643356643356,
|
|
"tf1": 0.9903614457831326,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"Image": 0.0,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.36363636363636365,
|
|
"H1": 1.0
|
|
},
|
|
"time_ms": 27.231791
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "SPARSE-2024-INV-1234_borderless_table",
|
|
"file_size": 19649,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.3963732861565679,
|
|
"tf1": 0.9819819819819819,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.5338345864661654,
|
|
"ListItem": 0.7058823529411764,
|
|
"H1": 1.0,
|
|
"Table": 0.0,
|
|
"H2": 0.0,
|
|
"Image": 0.0
|
|
},
|
|
"time_ms": 15.271958
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "WARN-Report-for-7-1-2015-to-03-25-2016",
|
|
"file_size": 478619,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.05281376360328497,
|
|
"tf1": 0.6955598455598456,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Table": 0.21081616918162793,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.0013166556945358788
|
|
},
|
|
"time_ms": 3635.487541
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "amt_handbook_sample",
|
|
"file_size": 673416,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.5228582619886968,
|
|
"tf1": 0.9296482412060302,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"H2": 1.0,
|
|
"Paragraph": 0.7057243578982709
|
|
},
|
|
"time_ms": 71.71195800000001
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "annotations-unicode-issues",
|
|
"file_size": 57927,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.75,
|
|
"tf1": 0.7058823529411764,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.75
|
|
},
|
|
"time_ms": 0.5435
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "code_and_formula",
|
|
"file_size": 89031,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.05883088284234227,
|
|
"tf1": 0.995068493150685,
|
|
"order_score": 0.5555555555555556,
|
|
"per_type_sf1": {
|
|
"Code": 0.0,
|
|
"Formula": 0.0,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.8824632426351341,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 40.457417
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "cupertino_usd_4-6-16",
|
|
"file_size": 80578,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.16298195631528967,
|
|
"tf1": 0.99581589958159,
|
|
"order_score": 0.9166666666666666,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Image": 0.0,
|
|
"H1": 0.0,
|
|
"ListItem": 0.6445584045584046,
|
|
"Paragraph": 0.6666666666666667
|
|
},
|
|
"time_ms": 115.72041700000001
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "docling",
|
|
"file_size": 5566575,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.28541699352310707,
|
|
"tf1": 0.5477201925800057,
|
|
"order_score": 0.9591836734693877,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.3458823529411765,
|
|
"Table": 0.0,
|
|
"Image": 1.0,
|
|
"H2": 0.0,
|
|
"H1": 1.0,
|
|
"Code": 0.0,
|
|
"H3": 0.0,
|
|
"Paragraph": 0.5874091516260027
|
|
},
|
|
"time_ms": 407.67479099999997
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "extra-attrs-example",
|
|
"file_size": 15170,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 1.0,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 1.0
|
|
},
|
|
"time_ms": 0.694667
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "federal-register-2020-17221",
|
|
"file_size": 713992,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.013074659124635218,
|
|
"tf1": 0.9365957209094463,
|
|
"order_score": 0.7142857142857143,
|
|
"per_type_sf1": {
|
|
"H3": 0.0,
|
|
"ListItem": 0.1111111111111111,
|
|
"Table": 0.0,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.06542027851975256,
|
|
"H1": 0.0,
|
|
"H4": 0.0
|
|
},
|
|
"time_ms": 1550.5107919999998
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "figure_structure",
|
|
"file_size": 24694,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.08750000000000001,
|
|
"tf1": 0.8437499999999999,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Table": 0.0,
|
|
"Paragraph": 0.7000000000000001
|
|
},
|
|
"time_ms": 1.448834
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "hello_structure",
|
|
"file_size": 2641,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.43750000000000006,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.43750000000000006
|
|
},
|
|
"time_ms": 0.582666
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-1114-dedupe-chars",
|
|
"file_size": 43483,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.08115942028985508,
|
|
"tf1": 0.7586206896551724,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Paragraph": 0.4057971014492754
|
|
},
|
|
"time_ms": 1.290542
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-1181",
|
|
"file_size": 13437,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.15277777777777776,
|
|
"tf1": 0.47058823529411764,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.611111111111111,
|
|
"Table": 0.0
|
|
},
|
|
"time_ms": 1.598875
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-1279-example",
|
|
"file_size": 37214,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.053901996370235926,
|
|
"tf1": 0.6572637517630465,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.3773139745916515
|
|
},
|
|
"time_ms": 147.92987499999998
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-140-example",
|
|
"file_size": 5251,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 9.166417
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-316-example",
|
|
"file_size": 138691,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.05628379839930781,
|
|
"tf1": 0.9936070651815725,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.0,
|
|
"H2": 0.0,
|
|
"Table": 0.0,
|
|
"Image": 0.0,
|
|
"H4": 0.0,
|
|
"H3": 0.13333333333333333,
|
|
"Paragraph": 0.7611940298507462,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 398.55616599999996
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-461-example",
|
|
"file_size": 154889,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.0,
|
|
"Table": 0.0
|
|
},
|
|
"time_ms": 38.217166
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-463-example",
|
|
"file_size": 88676,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.3329404861359749,
|
|
"tf1": 0.9459459459459459,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Paragraph": 0.5258692365835224,
|
|
"H1": 1.0,
|
|
"Code": 0.0,
|
|
"Table": 0.6,
|
|
"H3": 0.0
|
|
},
|
|
"time_ms": 393.85279099999997
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-466-example",
|
|
"file_size": 13569,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.07142857142857144,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"Paragraph": 0.28571428571428575
|
|
},
|
|
"time_ms": 6.412167
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-53-example",
|
|
"file_size": 133586,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.006081081081081081,
|
|
"tf1": 0.9784461152882206,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"H3": 0.0,
|
|
"Table": 0.0,
|
|
"Paragraph": 0.0972972972972973,
|
|
"H2": 0.0
|
|
},
|
|
"time_ms": 1692.980125
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-598-example",
|
|
"file_size": 504551,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.45981167608286255,
|
|
"tf1": 0.9684684684684683,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.6666666666666666,
|
|
"ListItem": 1.0,
|
|
"H3": 0.0,
|
|
"Paragraph": 0.2305084745762712,
|
|
"H1": 0.5
|
|
},
|
|
"time_ms": 10.447208
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-842-example",
|
|
"file_size": 287855,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.6511627906976744,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"Paragraph": 0.0,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 9.490625000000001
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-848",
|
|
"file_size": 71805,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.18181818181818182,
|
|
"tf1": 0.25101214574898784,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.18181818181818182
|
|
},
|
|
"time_ms": 80.34983299999999
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-905",
|
|
"file_size": 18810,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.4753805774278216,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 1.0,
|
|
"Paragraph": 0.2784251968503937,
|
|
"H2": 0.0
|
|
},
|
|
"time_ms": 17.051333
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-982-example",
|
|
"file_size": 1029061,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.022535368630924996,
|
|
"tf1": 0.9867206842223722,
|
|
"order_score": 0.8363636363636363,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.42817200398757493,
|
|
"Image": 0.0,
|
|
"H1": 0.0,
|
|
"ListItem": 0.0,
|
|
"H3": 0.0,
|
|
"H2": 0.0,
|
|
"Code": 0.0
|
|
},
|
|
"time_ms": 522.062917
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "issue-987-test",
|
|
"file_size": 8805,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 0.697542
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "la-precinct-bulletin-2014-p1",
|
|
"file_size": 20188,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.6368159203980099,
|
|
"tf1": 0.9732484076433121,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.6368159203980099
|
|
},
|
|
"time_ms": 48.965083
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "line-char-render-example",
|
|
"file_size": 13649,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 1.0,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 1.0
|
|
},
|
|
"time_ms": 0.79
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "masterformat_partial_numbering",
|
|
"file_size": 2114,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.13188882680408104,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.9232217876285672,
|
|
"H2": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 2.180583
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "mcid_example",
|
|
"file_size": 24694,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.08750000000000001,
|
|
"tf1": 0.8437499999999999,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.7000000000000001,
|
|
"Table": 0.0,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 1.292542
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "multi_page",
|
|
"file_size": 128322,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.49192701345415907,
|
|
"tf1": 1.0,
|
|
"order_score": 0.98,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.5300775181178955,
|
|
"H2": 0.9523809523809523,
|
|
"Paragraph": 0.5415183022361487,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 136.66370799999999
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nics-background-checks-2015-11",
|
|
"file_size": 90468,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.3027458492975735,
|
|
"tf1": 0.995659432387312,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.8000000000000002,
|
|
"Table": 0.0,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.43295019157088127
|
|
},
|
|
"time_ms": 948.7863749999999
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nics-background-checks-2015-11-rotated",
|
|
"file_size": 90415,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.29359903381642516,
|
|
"tf1": 0.995659432387312,
|
|
"order_score": 0.6666666666666666,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Paragraph": 0.32318840579710145,
|
|
"Table": 0.0,
|
|
"H1": 0.8000000000000002
|
|
},
|
|
"time_ms": 955.461375
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_001",
|
|
"file_size": 525343,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.16162275626906916,
|
|
"tf1": 0.8511904761904762,
|
|
"order_score": 0.85,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.0,
|
|
"H1": 0.0,
|
|
"H2": 0.6666666666666666,
|
|
"Image": 0.0,
|
|
"Paragraph": 0.4041657024456475,
|
|
"Table": 0.0,
|
|
"H3": 0.0
|
|
},
|
|
"time_ms": 55.814
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_002",
|
|
"file_size": 126909,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.5075651664517359,
|
|
"tf1": 0.9734513274336283,
|
|
"order_score": 0.9696969696969697,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.4767623228014963,
|
|
"Paragraph": 0.5691708537522152
|
|
},
|
|
"time_ms": 105.37629199999999
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_003",
|
|
"file_size": 205324,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.04929625916604013,
|
|
"tf1": 0.9982558139534883,
|
|
"order_score": 0.8620689655172413,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Paragraph": 0.3450738141622809,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 233.34270800000002
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_004",
|
|
"file_size": 63070,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.9500000000000001,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0,
|
|
"H2": 0.0,
|
|
"H1": 0.0,
|
|
"Image": 0.0
|
|
},
|
|
"time_ms": 3.28275
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_005",
|
|
"file_size": 932335,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.042208193293959965,
|
|
"tf1": 0.9184587813620072,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.0,
|
|
"H1": 0.0,
|
|
"Paragraph": 0.6331228994093995,
|
|
"H2": 0.0,
|
|
"H3": 0.0
|
|
},
|
|
"time_ms": 50.399541
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_006",
|
|
"file_size": 80325,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.09775977597759776,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"H1": 0.0,
|
|
"Paragraph": 0.8798379837983799
|
|
},
|
|
"time_ms": 66.95991699999999
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_007",
|
|
"file_size": 185508,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.09036709758054368,
|
|
"tf1": 0.9325292122125896,
|
|
"order_score": 0.9230769230769231,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.0,
|
|
"H3": 0.0,
|
|
"H1": 0.0,
|
|
"H2": 0.2727272727272727,
|
|
"Image": 0.0,
|
|
"Paragraph": 0.35496447037960804
|
|
},
|
|
"time_ms": 213.947792
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_008",
|
|
"file_size": 143492,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.06047819971870605,
|
|
"tf1": 0.999498997995992,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.5443037974683544
|
|
},
|
|
"time_ms": 114.41025
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_009",
|
|
"file_size": 498806,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.27328627174760495,
|
|
"tf1": 0.9285714285714286,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"Paragraph": 0.5179293592829252,
|
|
"ListItem": 0.4705882352941177,
|
|
"Table": 0.0,
|
|
"H3": 0.4,
|
|
"H1": 0.0,
|
|
"H2": 0.5333333333333333
|
|
},
|
|
"time_ms": 131.397375
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_010",
|
|
"file_size": 107951,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.08553003586666215,
|
|
"tf1": 0.9604573606479276,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"H3": 0.0,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.680210811972535,
|
|
"ListItem": 0.3013698630136986
|
|
},
|
|
"time_ms": 74.780875
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_011",
|
|
"file_size": 191109,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.7251908396946565,
|
|
"tf1": 0.9900773963087915,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.7251908396946565
|
|
},
|
|
"time_ms": 408.90912499999996
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_012",
|
|
"file_size": 141727,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.18939196858330032,
|
|
"tf1": 1.0,
|
|
"order_score": 0.8823529411764706,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.6220400763793985,
|
|
"ListItem": 0.35185185185185186,
|
|
"H2": 0.0
|
|
},
|
|
"time_ms": 142.50954199999998
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_013",
|
|
"file_size": 61858,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.5628017815517815,
|
|
"tf1": 0.9879931389365352,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.7536213786213786,
|
|
"ListItem": 1.0,
|
|
"H1": 1.0
|
|
},
|
|
"time_ms": 56.554874999999996
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_014",
|
|
"file_size": 848830,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0,
|
|
"Table": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 60002.154833
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_015",
|
|
"file_size": 46652,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 60002.145874999995
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_016",
|
|
"file_size": 454010,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 60001.561541999996
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_017",
|
|
"file_size": 911122,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 60002.142083000006
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_018",
|
|
"file_size": 516747,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0,
|
|
"Table": 0.0
|
|
},
|
|
"time_ms": 60001.912791
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_019",
|
|
"file_size": 183646,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0,
|
|
"ListItem": 0.0,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 60002.154875
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_020",
|
|
"file_size": 115368,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 60002.132417
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_021",
|
|
"file_size": 469199,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 60002.129792
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_022",
|
|
"file_size": 34841,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0,
|
|
"ListItem": 0.0,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 60002.15825
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_023",
|
|
"file_size": 74186,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 60002.682667
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_024",
|
|
"file_size": 46310,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.06791111111111112,
|
|
"tf1": 0.9824561403508771,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Image": 0.0,
|
|
"Paragraph": 0.6791111111111111,
|
|
"H2": 0.0
|
|
},
|
|
"time_ms": 39578.7155
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_025",
|
|
"file_size": 2094,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.9041095890410958,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"Code": 0.0,
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 5.671084
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_026",
|
|
"file_size": 44245,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.03920982761134892,
|
|
"tf1": 1.0,
|
|
"order_score": 0.7307692307692307,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Table": 0.0,
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.3920982761134892
|
|
},
|
|
"time_ms": 142.219333
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_027",
|
|
"file_size": 390183,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.16518735355445174,
|
|
"tf1": 0.9637832984136485,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.5714285714285715,
|
|
"H1": 0.0,
|
|
"Paragraph": 0.026908664048038287,
|
|
"Table": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 434.814417
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_028",
|
|
"file_size": 269676,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.3037978045496179,
|
|
"tf1": 0.7430901676483914,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"Paragraph": 0.20697716336734534,
|
|
"H2": 0.3333333333333333,
|
|
"ListItem": 0.0,
|
|
"H1": 0.5263157894736842
|
|
},
|
|
"time_ms": 179.656958
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_029",
|
|
"file_size": 274635,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.09611398300638918,
|
|
"tf1": 0.9741992882562277,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Paragraph": 0.7311007420564593,
|
|
"Image": 0.0,
|
|
"H1": 0.0,
|
|
"Code": 0.0,
|
|
"H3": 0.0,
|
|
"ListItem": 0.5475324675324675
|
|
},
|
|
"time_ms": 420.526041
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_030",
|
|
"file_size": 772983,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.012763906790673431,
|
|
"tf1": 0.9462672372800761,
|
|
"order_score": 0.8333333333333334,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"H1": 0.0,
|
|
"Paragraph": 0.2425142290227952,
|
|
"Image": 0.0,
|
|
"H3": 0.0,
|
|
"ListItem": 0.0,
|
|
"H2": 0.0
|
|
},
|
|
"time_ms": 174.994333
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_031",
|
|
"file_size": 40157,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.09835564628235795,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.6884895239765056,
|
|
"ListItem": 0.0,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 38.75725
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_032",
|
|
"file_size": 112154,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.49853535022589773,
|
|
"tf1": 0.9598278335724533,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.49853535022589773
|
|
},
|
|
"time_ms": 121.809584
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_033",
|
|
"file_size": 181946,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.025481842263849183,
|
|
"tf1": 0.9208443271767809,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.280300264902341,
|
|
"H2": 0.0,
|
|
"Image": 0.0
|
|
},
|
|
"time_ms": 132.925792
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_034",
|
|
"file_size": 219515,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.0,
|
|
"Table": 0.0
|
|
},
|
|
"time_ms": 60001.203457999996
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_035",
|
|
"file_size": 813601,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 60002.132625
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_036",
|
|
"file_size": 399271,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.47552837240924833,
|
|
"tf1": 0.9928288822947576,
|
|
"order_score": 0.9714285714285714,
|
|
"per_type_sf1": {
|
|
"H2": 0.8761904761904763,
|
|
"Image": 0.0,
|
|
"H1": 0.6666666666666666,
|
|
"H3": 0.0,
|
|
"Paragraph": 0.48596864230090453
|
|
},
|
|
"time_ms": 6289.424
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_037",
|
|
"file_size": 561676,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.3409113872039522,
|
|
"tf1": 0.9838945827232797,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.9090909090909091,
|
|
"Image": 0.0,
|
|
"Paragraph": 0.2871311496186737,
|
|
"H2": 0.0,
|
|
"ListItem": 0.08372093023255814
|
|
},
|
|
"time_ms": 41.627625
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_038",
|
|
"file_size": 2391915,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.21987543906670995,
|
|
"tf1": 0.924114671163575,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"Table": 0.7804878048780488,
|
|
"H1": 0.0,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.5169172932330827
|
|
},
|
|
"time_ms": 20.19625
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_039",
|
|
"file_size": 69496,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.19055114231821405,
|
|
"tf1": 0.9329268292682927,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.1878787878787878,
|
|
"Table": 0.0,
|
|
"Image": 0.0,
|
|
"ListItem": 0.9540918888107833,
|
|
"H2": 0.0
|
|
},
|
|
"time_ms": 27.381249999999998
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_040",
|
|
"file_size": 476639,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.03826981135490284,
|
|
"tf1": 0.9217075984835997,
|
|
"order_score": 0.6875,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Paragraph": 0.34151319553905424,
|
|
"H2": 0.06593406593406594,
|
|
"ListItem": 0.11834319526627218,
|
|
"Code": 0.0,
|
|
"Table": 0.0,
|
|
"Image": 0.0,
|
|
"H3": 0.0
|
|
},
|
|
"time_ms": 1362.8575
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_041",
|
|
"file_size": 733602,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.1683106139409614,
|
|
"tf1": 0.8784194528875381,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"ListItem": 0.8749999999999999,
|
|
"Paragraph": 0.26972736729153707,
|
|
"H1": 0.0,
|
|
"Image": 0.0
|
|
},
|
|
"time_ms": 144.727833
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_042",
|
|
"file_size": 166420,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.40861620716691066,
|
|
"tf1": 0.9983779399837793,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.858414642472381,
|
|
"H1": 0.9090909090909091
|
|
},
|
|
"time_ms": 76.143583
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_043",
|
|
"file_size": 119114,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.09561729524838837,
|
|
"tf1": 0.9969707169303265,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.2517902477322719,
|
|
"H2": 0.0,
|
|
"H1": 0.0,
|
|
"ListItem": 0.4
|
|
},
|
|
"time_ms": 69.32575
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_044",
|
|
"file_size": 107812,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.9131534569983136,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 92.623917
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_045",
|
|
"file_size": 129544,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.9493087557603688,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Code": 0.0,
|
|
"Paragraph": 0.0,
|
|
"H1": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 130.00858300000002
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_046",
|
|
"file_size": 594787,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.12668543433046067,
|
|
"tf1": 0.9220430107526881,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"H2": 0.0,
|
|
"Image": 0.0,
|
|
"ListItem": 0.7777777777777778,
|
|
"Table": 0.0,
|
|
"Paragraph": 0.34472595940135425
|
|
},
|
|
"time_ms": 57.72975
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_047",
|
|
"file_size": 420052,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.02812805825444189,
|
|
"tf1": 0.9607629427792915,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.4500489320710702,
|
|
"H1": 0.0,
|
|
"Image": 0.0,
|
|
"H2": 0.0,
|
|
"H3": 0.0
|
|
},
|
|
"time_ms": 267.176875
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_048",
|
|
"file_size": 572026,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.35265164497062435,
|
|
"tf1": 0.9872572815533981,
|
|
"order_score": 0.7619047619047619,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.23095114514760068,
|
|
"Image": 0.0,
|
|
"H1": 1.0,
|
|
"Table": 0.0,
|
|
"ListItem": 0.5294117647058824,
|
|
"H2": 0.0
|
|
},
|
|
"time_ms": 478.84608299999996
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_049",
|
|
"file_size": 122803,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.03938708350473056,
|
|
"tf1": 0.9187279151943463,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Paragraph": 0.3938708350473056,
|
|
"H2": 0.0,
|
|
"Image": 0.0
|
|
},
|
|
"time_ms": 43.9495
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "nougat_050",
|
|
"file_size": 108353,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.088240296342817,
|
|
"tf1": 0.9424141749723145,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"H1": 0.0,
|
|
"Paragraph": 0.7941626670853531
|
|
},
|
|
"time_ms": 74.518708
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "page-boxes-example",
|
|
"file_size": 1187,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 1.0,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 1.0
|
|
},
|
|
"time_ms": 0.9064169999999999
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_embedded",
|
|
"file_size": 109798,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.018744142455482657,
|
|
"tf1": 0.9727626459143969,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"Table": 0.0,
|
|
"H1": 0.0,
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.20618556701030924
|
|
},
|
|
"time_ms": 20.805125
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_google_docs",
|
|
"file_size": 80100,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.3692307692307692,
|
|
"tf1": 0.93841642228739,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"Table": 0.0,
|
|
"H2": 0.0,
|
|
"H1": 1.0,
|
|
"Paragraph": 0.8
|
|
},
|
|
"time_ms": 426.252625
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_image_only_german",
|
|
"file_size": 189884,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.017077798861480076,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"H3": 0.0,
|
|
"Image": 0.0,
|
|
"H1": 0.0,
|
|
"Paragraph": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 0.6677500000000001
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_large_tables",
|
|
"file_size": 2636957,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.23737343577541425,
|
|
"tf1": 0.6043638697256537,
|
|
"order_score": 0.9166666666666666,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Paragraph": 0.392231319988672,
|
|
"Code": 0.0,
|
|
"Table": 0.0,
|
|
"H1": 0.9333333333333333,
|
|
"Image": 0.0,
|
|
"ListItem": 0.07357859531772576
|
|
},
|
|
"time_ms": 925.44175
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_medium",
|
|
"file_size": 358744,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.4312688552103267,
|
|
"tf1": 0.960789030550878,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 1.0,
|
|
"H2": 0.11764705882352941,
|
|
"Code": 0.0,
|
|
"Paragraph": 0.7046380272298027
|
|
},
|
|
"time_ms": 314.4405
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_medium_tables",
|
|
"file_size": 245260,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.16719819057094618,
|
|
"tf1": 0.5716484926234766,
|
|
"order_score": 0.8,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.3230019590795889,
|
|
"Table": 0.0,
|
|
"Image": 0.0,
|
|
"H1": 0.3870967741935484,
|
|
"Paragraph": 0.3135818436308216,
|
|
"H2": 0.0
|
|
},
|
|
"time_ms": 739.500042
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_non_searchable",
|
|
"file_size": 70489,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.06060606060606061,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"H1": 0.0,
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 0.636541
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_ocr_rotated",
|
|
"file_size": 94703,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 0.436209
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_ocr_rotated_270",
|
|
"file_size": 94702,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 0.393042
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_ocr_rotated_90",
|
|
"file_size": 94126,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 0.394458
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_ocr_test",
|
|
"file_size": 93549,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 0.3265
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_protected",
|
|
"file_size": 270117,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.33068260158934476,
|
|
"tf1": 0.6205607476635514,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Paragraph": 0.613073248439102,
|
|
"Image": 0.0,
|
|
"H1": 0.7058823529411764,
|
|
"ListItem": 0.7618181818181818,
|
|
"Code": 0.0
|
|
},
|
|
"time_ms": 81.8295
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_scanned_ocr",
|
|
"file_size": 69966,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.030703960682749627,
|
|
"tf1": 0.9090909090909091,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H4": 0.0,
|
|
"Paragraph": 0.5833752529722429,
|
|
"H1": 0.0,
|
|
"H3": 0.0,
|
|
"H2": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 334.547375
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_structure",
|
|
"file_size": 34467,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.31331569664903003,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H3": 0.0,
|
|
"H2": 0.5,
|
|
"Table": 0.0,
|
|
"Paragraph": 0.5000000000000001,
|
|
"ListItem": 0.7698412698412698,
|
|
"H1": 0.4000000000000001
|
|
},
|
|
"time_ms": 7.093666
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_tables",
|
|
"file_size": 23572,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.6868055555555556,
|
|
"tf1": 0.6666666666666666,
|
|
"order_score": 0.5,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.6222222222222222,
|
|
"Table": 0.7083333333333334
|
|
},
|
|
"time_ms": 12.932542
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdf_tiny_memo",
|
|
"file_size": 13374,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.8019323671497585,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.8695652173913044,
|
|
"Paragraph": 0.6666666666666666
|
|
},
|
|
"time_ms": 1.3340839999999998
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_001",
|
|
"file_size": 44245,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0439904791188115,
|
|
"tf1": 1.0,
|
|
"order_score": 0.76,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.43990479118811504,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 138.5075
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_002",
|
|
"file_size": 390183,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.16518735355445174,
|
|
"tf1": 0.9637832984136485,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Table": 0.0,
|
|
"H2": 0.5714285714285715,
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.026908664048038287
|
|
},
|
|
"time_ms": 426.770334
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_003",
|
|
"file_size": 269676,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.36837507227594307,
|
|
"tf1": 0.7430901676483914,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"H1": 0.5263157894736842,
|
|
"Paragraph": 0.2451542315313605,
|
|
"H2": 0.3333333333333333
|
|
},
|
|
"time_ms": 177.676
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_004",
|
|
"file_size": 274635,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.09660772981339964,
|
|
"tf1": 0.9741992882562277,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Code": 0.0,
|
|
"Image": 0.0,
|
|
"H1": 0.0,
|
|
"H3": 0.0,
|
|
"ListItem": 0.5475324675324675,
|
|
"Paragraph": 0.740481931389658
|
|
},
|
|
"time_ms": 424.188209
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_005",
|
|
"file_size": 772983,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0021367521367521365,
|
|
"tf1": 0.9462672372800761,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H3": 0.0,
|
|
"H1": 0.0,
|
|
"Paragraph": 0.034188034188034185,
|
|
"H2": 0.0,
|
|
"Image": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 175.913792
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_006",
|
|
"file_size": 40157,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.09835564628235795,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.6884895239765056,
|
|
"H1": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 36.73925
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_007",
|
|
"file_size": 112154,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.6176248381503184,
|
|
"tf1": 0.9598278335724533,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.6176248381503184
|
|
},
|
|
"time_ms": 121.78358300000001
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_008",
|
|
"file_size": 181946,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.03715361006382678,
|
|
"tf1": 0.9208443271767809,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Table": 0.0,
|
|
"Image": 0.0,
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.4086897107020946
|
|
},
|
|
"time_ms": 132.928292
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_009",
|
|
"file_size": 219515,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.0,
|
|
"Table": 0.0,
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 60002.517208
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_010",
|
|
"file_size": 813601,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.0,
|
|
"Table": 0.0
|
|
},
|
|
"time_ms": 60002.135125
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_011",
|
|
"file_size": 399271,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.47552837240924833,
|
|
"tf1": 0.9928288822947576,
|
|
"order_score": 0.9714285714285714,
|
|
"per_type_sf1": {
|
|
"H2": 0.8761904761904763,
|
|
"H1": 0.6666666666666666,
|
|
"Image": 0.0,
|
|
"Paragraph": 0.48596864230090453,
|
|
"H3": 0.0
|
|
},
|
|
"time_ms": 6590.070334
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_012",
|
|
"file_size": 561676,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.3409113872039522,
|
|
"tf1": 0.9838945827232797,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"H1": 0.9090909090909091,
|
|
"Image": 0.0,
|
|
"Paragraph": 0.2871311496186737,
|
|
"ListItem": 0.08372093023255814
|
|
},
|
|
"time_ms": 42.364
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_013",
|
|
"file_size": 2391915,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.22230802921859835,
|
|
"tf1": 0.924114671163575,
|
|
"order_score": 0.9,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"Table": 0.7804878048780488,
|
|
"H1": 0.0,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.5485409652076318
|
|
},
|
|
"time_ms": 20.023540999999998
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_014",
|
|
"file_size": 69496,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.1898085384135548,
|
|
"tf1": 0.9329268292682927,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.1797101449275362,
|
|
"Table": 0.0,
|
|
"ListItem": 0.9540918888107833
|
|
},
|
|
"time_ms": 27.554333
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_015",
|
|
"file_size": 476639,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.04486459316508731,
|
|
"tf1": 0.9217075984835997,
|
|
"order_score": 0.6666666666666666,
|
|
"per_type_sf1": {
|
|
"H3": 0.0,
|
|
"Table": 0.0,
|
|
"H1": 0.0,
|
|
"Code": 0.0,
|
|
"H2": 0.10364372469635627,
|
|
"ListItem": 0.1098901098901099,
|
|
"Paragraph": 0.352665931066276,
|
|
"Image": 0.0
|
|
},
|
|
"time_ms": 1362.372959
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_016",
|
|
"file_size": 733602,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.1683106139409614,
|
|
"tf1": 0.8784194528875381,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.8749999999999999,
|
|
"Paragraph": 0.26972736729153707,
|
|
"Image": 0.0,
|
|
"H2": 0.0,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 144.69341699999998
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_017",
|
|
"file_size": 166420,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.466680906889572,
|
|
"tf1": 0.9983779399837793,
|
|
"order_score": 0.9583333333333334,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.5880354303307467,
|
|
"H1": 0.9090909090909091,
|
|
"ListItem": 0.0,
|
|
"H2": 0.22727272727272727
|
|
},
|
|
"time_ms": 75.983834
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_018",
|
|
"file_size": 119114,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.09610431700415677,
|
|
"tf1": 0.9969707169303265,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.2571474870457245,
|
|
"ListItem": 0.4,
|
|
"H1": 0.0,
|
|
"H2": 0.0
|
|
},
|
|
"time_ms": 68.253916
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_019",
|
|
"file_size": 107812,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.9131534569983136,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 90.56441699999999
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_020",
|
|
"file_size": 129544,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.9493087557603688,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Code": 0.0,
|
|
"Paragraph": 0.0,
|
|
"H1": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 129.024125
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_021",
|
|
"file_size": 594787,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.09109407801099663,
|
|
"tf1": 0.9220430107526881,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"ListItem": 0.707070707070707,
|
|
"H1": 0.0,
|
|
"H3": 0.0,
|
|
"Paragraph": 0.3166460680675221,
|
|
"Table": 0.0,
|
|
"Image": 0.0
|
|
},
|
|
"time_ms": 58.513792
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_022",
|
|
"file_size": 420052,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.039481403582141604,
|
|
"tf1": 0.9607629427792915,
|
|
"order_score": 0.9743589743589743,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.0,
|
|
"H2": 0.0,
|
|
"Image": 0.0,
|
|
"Paragraph": 0.6317024573142657,
|
|
"H3": 0.0,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 268.217542
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_023",
|
|
"file_size": 572026,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.35265164497062435,
|
|
"tf1": 0.9872572815533981,
|
|
"order_score": 0.7619047619047619,
|
|
"per_type_sf1": {
|
|
"H1": 1.0,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.23095114514760068,
|
|
"ListItem": 0.5294117647058824,
|
|
"Image": 0.0,
|
|
"Table": 0.0
|
|
},
|
|
"time_ms": 478.350708
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_024",
|
|
"file_size": 122803,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.03938708350473056,
|
|
"tf1": 0.9187279151943463,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Image": 0.0,
|
|
"Paragraph": 0.3938708350473056,
|
|
"H2": 0.0
|
|
},
|
|
"time_ms": 43.805709
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_025",
|
|
"file_size": 108353,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.088240296342817,
|
|
"tf1": 0.9424141749723145,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.7941626670853531,
|
|
"H2": 0.0,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 74.73783300000001
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_026",
|
|
"file_size": 767342,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.022355370322169704,
|
|
"tf1": 0.9694385216773276,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.35768592515471526,
|
|
"ListItem": 0.0,
|
|
"Image": 0.0,
|
|
"H2": 0.0,
|
|
"H3": 0.0,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 648.429625
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_027",
|
|
"file_size": 163606,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.05758582502768549,
|
|
"tf1": 0.893852175915266,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Code": 0.0,
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.8062015503875969,
|
|
"H3": 0.0,
|
|
"H2": 0.0
|
|
},
|
|
"time_ms": 248.80966700000002
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_028",
|
|
"file_size": 142067,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.05120511633427794,
|
|
"tf1": 0.9889549702633814,
|
|
"order_score": 0.9411764705882353,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.0,
|
|
"Image": 0.0,
|
|
"H2": 0.0,
|
|
"H1": 0.0,
|
|
"Paragraph": 0.6144613960113353
|
|
},
|
|
"time_ms": 374.780791
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_029",
|
|
"file_size": 5262991,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.4705867950896485,
|
|
"tf1": 0.9899732620320856,
|
|
"order_score": 0.9122807017543859,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"H2": 0.5273368606701939,
|
|
"Paragraph": 0.28203452132873474,
|
|
"ListItem": 0.5559322033898306,
|
|
"H1": 0.888888888888889,
|
|
"Table": 0.0
|
|
},
|
|
"time_ms": 328.02725
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_030",
|
|
"file_size": 110445,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.3263481888481889,
|
|
"tf1": 0.9885496183206107,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.5714285714285715,
|
|
"Image": 0.0,
|
|
"ListItem": 0.6444444444444444,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.3415750915750916
|
|
},
|
|
"time_ms": 15.219375
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_031",
|
|
"file_size": 4873,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.15949820788530467,
|
|
"tf1": 0.9797979797979798,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.0,
|
|
"Paragraph": 0.478494623655914
|
|
},
|
|
"time_ms": 2.805917
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_032",
|
|
"file_size": 108342,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.8148037399592278,
|
|
"tf1": 0.9808695652173912,
|
|
"order_score": 0.9523809523809523,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.874018699796138,
|
|
"H1": 0.8000000000000002
|
|
},
|
|
"time_ms": 79.954084
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_033",
|
|
"file_size": 11564,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.3166666666666667,
|
|
"tf1": 0.7258064516129032,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.3166666666666667
|
|
},
|
|
"time_ms": 438.180292
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_034",
|
|
"file_size": 27491,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.4982746721877157,
|
|
"tf1": 0.9672619047619048,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.8333333333333334,
|
|
"Paragraph": 0.15458937198067635,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 3.718417
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_035",
|
|
"file_size": 79182,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.12037037037037039,
|
|
"tf1": 0.6153846153846154,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"Paragraph": 0.24074074074074078
|
|
},
|
|
"time_ms": 6.707249999999999
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_036",
|
|
"file_size": 409867,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.20641550739228032,
|
|
"tf1": 0.9398301956441492,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"H2": 0.0,
|
|
"Image": 0.0,
|
|
"H1": 0.6666666666666666,
|
|
"Paragraph": 0.4295659442175381,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 507.03916699999996
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_037",
|
|
"file_size": 282613,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.16750443764060138,
|
|
"tf1": 0.9476351351351352,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"Paragraph": 0.33500887528120277
|
|
},
|
|
"time_ms": 67.149959
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_038",
|
|
"file_size": 14168,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.05121544244605145,
|
|
"tf1": 0.9472329472329473,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Paragraph": 0.563369866906566,
|
|
"H1": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 20.253334000000002
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_039",
|
|
"file_size": 69411,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.96617050067659,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Paragraph": 0.0,
|
|
"Table": 0.0,
|
|
"Code": 0.0
|
|
},
|
|
"time_ms": 478.314625
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_040",
|
|
"file_size": 108450,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.012580486684088715,
|
|
"tf1": 0.9312906220984216,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.06290243342044358,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 91.420083
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_041",
|
|
"file_size": 41518,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.2585780736069753,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.5925925925925926,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.47398843930635837,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 32.957417
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_042",
|
|
"file_size": 197955,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.2478414973158646,
|
|
"tf1": 0.9533116178067318,
|
|
"order_score": 0.8095238095238095,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.18181818181818182,
|
|
"Table": 0.0,
|
|
"Image": 0.0,
|
|
"H2": 0.14857142857142858,
|
|
"H3": 0.45454545454545453,
|
|
"Paragraph": 0.2575838754943233,
|
|
"H4": 0.0,
|
|
"H1": 0.6666666666666666
|
|
},
|
|
"time_ms": 29.580958000000003
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_043",
|
|
"file_size": 106715,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.01598806786932965,
|
|
"tf1": 0.8796331677493312,
|
|
"order_score": 0.75,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.2238329501706151,
|
|
"H1": 0.0,
|
|
"ListItem": 0.0
|
|
},
|
|
"time_ms": 117.06741699999999
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_044",
|
|
"file_size": 143107,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.03231158439630513,
|
|
"tf1": 0.954349698535745,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.484673765944577,
|
|
"H1": 0.0,
|
|
"H2": 0.0,
|
|
"Image": 0.0,
|
|
"ListItem": 0.0,
|
|
"Table": 0.0
|
|
},
|
|
"time_ms": 178.23325
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_045",
|
|
"file_size": 19551,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.5666103155731371,
|
|
"tf1": 0.8110760602874167,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.5816568243965754,
|
|
"ListItem": 0.0,
|
|
"H1": 0.8461538461538461
|
|
},
|
|
"time_ms": 92.03658300000001
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_046",
|
|
"file_size": 246086,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.03316347421409385,
|
|
"tf1": 0.9862903225806451,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.16581737107046926,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 68.52829200000001
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_047",
|
|
"file_size": 130632,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.3947751776903112,
|
|
"tf1": 0.9960132890365448,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"Paragraph": 0.8601858209976903,
|
|
"H1": 0.8333333333333333,
|
|
"H2": 0.0,
|
|
"ListItem": 0.6666666666666666
|
|
},
|
|
"time_ms": 32.927083
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_048",
|
|
"file_size": 640060,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.17416208349184967,
|
|
"tf1": 0.9911868390129259,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H3": 0.0,
|
|
"Paragraph": 0.7865933358695947,
|
|
"H1": 0.0,
|
|
"ListItem": 1.0,
|
|
"H2": 0.0,
|
|
"Image": 0.0
|
|
},
|
|
"time_ms": 85.24920800000001
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_049",
|
|
"file_size": 152698,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.3529674986998439,
|
|
"tf1": 0.9579256360078278,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.0,
|
|
"H1": 0.0,
|
|
"Paragraph": 0.7696343556169827,
|
|
"H2": 0.7782520325203252
|
|
},
|
|
"time_ms": 151.932459
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdfa_050",
|
|
"file_size": 308633,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.1638305374431484,
|
|
"tf1": 0.9950358891795801,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"ListItem": 0.2,
|
|
"H2": 0.0,
|
|
"Paragraph": 0.7468137621020389
|
|
},
|
|
"time_ms": 1456.534042
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pdffill-demo",
|
|
"file_size": 108017,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.01429490715205001,
|
|
"tf1": 0.848816029143898,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"Paragraph": 0.20012870012870015,
|
|
"H2": 0.0,
|
|
"H3": 0.0,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 9.011625
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "picture_classification",
|
|
"file_size": 212855,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.35926971326164875,
|
|
"tf1": 0.9931506849315068,
|
|
"order_score": 0.6666666666666666,
|
|
"per_type_sf1": {
|
|
"H2": 0.0,
|
|
"Image": 0.0,
|
|
"ListItem": 0.0,
|
|
"H1": 1.0,
|
|
"Paragraph": 0.31123655913978493
|
|
},
|
|
"time_ms": 15.800999999999998
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pr-138-example",
|
|
"file_size": 89950,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.3643100234277329,
|
|
"tf1": 0.9921304103428893,
|
|
"order_score": 0.9523809523809523,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"H1": 0.6666666666666666,
|
|
"ListItem": 0.32000000000000006,
|
|
"Paragraph": 0.336433567610662
|
|
},
|
|
"time_ms": 413.28695799999997
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "pr-88-example",
|
|
"file_size": 95332,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.05226480836236934,
|
|
"tf1": 0.9259259259259259,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Image": 0.0,
|
|
"H3": 0.0,
|
|
"Paragraph": 0.7317073170731707,
|
|
"H1": 0.0,
|
|
"H2": 0.0
|
|
},
|
|
"time_ms": 4.299291
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "redp5110_sampled",
|
|
"file_size": 1277938,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.2641690544412553,
|
|
"tf1": 0.9707210975405722,
|
|
"order_score": 0.8783783783783784,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"H3": 0.09380863039399624,
|
|
"ListItem": 0.3318681318681318,
|
|
"Paragraph": 0.5274282862972106,
|
|
"H1": 0.8181818181818182,
|
|
"Code": 0.0,
|
|
"H2": 0.24314821334272135,
|
|
"Image": 0.0
|
|
},
|
|
"time_ms": 313.842625
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "right_to_left_02",
|
|
"file_size": 92036,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.8042328042328042,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.0,
|
|
"Image": 0.0
|
|
},
|
|
"time_ms": 23.100125
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "right_to_left_03",
|
|
"file_size": 278131,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.07179487179487179,
|
|
"tf1": 0.8467741935483871,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"H2": 0.2333333333333333,
|
|
"Table": 0.0,
|
|
"Paragraph": 0.0,
|
|
"Image": 0.0
|
|
},
|
|
"time_ms": 27.722
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "scotus-transcript-p1",
|
|
"file_size": 84509,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.23952095808383228,
|
|
"tf1": 0.8369565217391304,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.23952095808383228
|
|
},
|
|
"time_ms": 7.392791
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "senate-expenditures",
|
|
"file_size": 53481,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.0,
|
|
"tf1": 0.012852199703410776,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Table": 0.0,
|
|
"Paragraph": 0.0
|
|
},
|
|
"time_ms": 217.637792
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "table-curves-example",
|
|
"file_size": 153741,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.1509433962264151,
|
|
"tf1": 0.998003992015968,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.6037735849056604,
|
|
"Table": 0.0
|
|
},
|
|
"time_ms": 38.346958
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "test",
|
|
"file_size": 92971,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.08058714103011276,
|
|
"tf1": 0.9973684210526316,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 0.4029357051505638,
|
|
"H1": 0.0
|
|
},
|
|
"time_ms": 65.131709
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "test-punkt",
|
|
"file_size": 14806,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 1.0,
|
|
"tf1": 1.0,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"Paragraph": 1.0
|
|
},
|
|
"time_ms": 0.86975
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"name": "word365_structure",
|
|
"file_size": 99253,
|
|
"results": [
|
|
{
|
|
"pipeline": "baseline",
|
|
"sf1": 0.24705882352941178,
|
|
"tf1": 0.7727272727272727,
|
|
"order_score": 1.0,
|
|
"per_type_sf1": {
|
|
"H1": 0.0,
|
|
"Paragraph": 0.0,
|
|
"ListItem": 0.0,
|
|
"Table": 0.8235294117647058
|
|
},
|
|
"time_ms": 3.985791
|
|
}
|
|
]
|
|
}
|
|
]
|
|
}
|