668 lines
10 KiB
JSON
668 lines
10 KiB
JSON
{
|
|
"artifact": "figure-label-audit",
|
|
"curator": "sphinx",
|
|
"png": "/root/troupe-standup/wizard-work/sphinx/transformer-archive/figure1.png",
|
|
"png_bytes": 159152,
|
|
"manifest": "/root/troupe-standup/wizard-work/sphinx/transformer-archive/manifest.json",
|
|
"source_snapshot": "/root/troupe-standup/wizard-work/sphinx/transformer-archive/SOURCE-text.md",
|
|
"ocr_engine": "/usr/bin/tesseract",
|
|
"ocr_psm": "11",
|
|
"ocr_word_count_raw": 69,
|
|
"ocr_word_count_joined": 67,
|
|
"ocr_group_count": 40,
|
|
"labels_graded": 13,
|
|
"ocr_exact": 6,
|
|
"ocr_boundary": 7,
|
|
"ocr_absent": 0,
|
|
"faults": [
|
|
{
|
|
"label": "Inputs",
|
|
"ocr_status": "OCR_BOUNDARY"
|
|
},
|
|
{
|
|
"label": "Input Embedding",
|
|
"ocr_status": "OCR_BOUNDARY"
|
|
},
|
|
{
|
|
"label": "Positional Encoding",
|
|
"ocr_status": "OCR_BOUNDARY"
|
|
},
|
|
{
|
|
"label": "Feed Forward",
|
|
"ocr_status": "OCR_BOUNDARY"
|
|
},
|
|
{
|
|
"label": "Output Embedding",
|
|
"ocr_status": "OCR_BOUNDARY"
|
|
},
|
|
{
|
|
"label": "Masked Multi-Head Attention",
|
|
"ocr_status": "OCR_BOUNDARY"
|
|
},
|
|
{
|
|
"label": "Output Probabilities",
|
|
"ocr_status": "OCR_BOUNDARY"
|
|
}
|
|
],
|
|
"rows": [
|
|
{
|
|
"label": "Inputs",
|
|
"ocr_status": "OCR_BOUNDARY",
|
|
"present_in_text": false,
|
|
"tokens": [
|
|
"inputs"
|
|
],
|
|
"ocr_run": []
|
|
},
|
|
{
|
|
"label": "Input Embedding",
|
|
"ocr_status": "OCR_BOUNDARY",
|
|
"present_in_text": false,
|
|
"tokens": [
|
|
"input",
|
|
"embedding"
|
|
],
|
|
"ocr_run": []
|
|
},
|
|
{
|
|
"label": "Positional Encoding",
|
|
"ocr_status": "OCR_BOUNDARY",
|
|
"present_in_text": true,
|
|
"tokens": [
|
|
"positional",
|
|
"encoding"
|
|
],
|
|
"ocr_run": []
|
|
},
|
|
{
|
|
"label": "Nx",
|
|
"ocr_status": "OCR_EXACT",
|
|
"present_in_text": false,
|
|
"tokens": [
|
|
"nx"
|
|
],
|
|
"ocr_run": [
|
|
{
|
|
"text": "Nx",
|
|
"x": 1387,
|
|
"y": 1045
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"label": "Multi-Head Attention",
|
|
"ocr_status": "OCR_EXACT",
|
|
"present_in_text": true,
|
|
"tokens": [
|
|
"multi",
|
|
"head",
|
|
"attention"
|
|
],
|
|
"ocr_run": [
|
|
{
|
|
"text": "Multi-Head",
|
|
"x": 908,
|
|
"y": 928
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"label": "Add & Norm",
|
|
"ocr_status": "OCR_EXACT",
|
|
"present_in_text": false,
|
|
"tokens": [
|
|
"add",
|
|
"norm"
|
|
],
|
|
"ocr_run": [
|
|
{
|
|
"text": "|",
|
|
"x": 845,
|
|
"y": 511
|
|
},
|
|
{
|
|
"text": "}<s",
|
|
"x": 1199,
|
|
"y": 511
|
|
},
|
|
{
|
|
"text": "Add",
|
|
"x": 887,
|
|
"y": 521
|
|
},
|
|
{
|
|
"text": "&",
|
|
"x": 998,
|
|
"y": 521
|
|
},
|
|
{
|
|
"text": "Norm",
|
|
"x": 1047,
|
|
"y": 521
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"label": "Feed Forward",
|
|
"ocr_status": "OCR_BOUNDARY",
|
|
"present_in_text": false,
|
|
"tokens": [
|
|
"feed",
|
|
"forward"
|
|
],
|
|
"ocr_run": []
|
|
},
|
|
{
|
|
"label": "Outputs (shifted right)",
|
|
"ocr_status": "OCR_EXACT",
|
|
"present_in_text": false,
|
|
"tokens": [
|
|
"outputs",
|
|
"shifted",
|
|
"right"
|
|
],
|
|
"ocr_run": [
|
|
{
|
|
"text": "(shifted",
|
|
"x": 862,
|
|
"y": 2175
|
|
},
|
|
{
|
|
"text": "right)",
|
|
"x": 1071,
|
|
"y": 2175
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"label": "Output Embedding",
|
|
"ocr_status": "OCR_BOUNDARY",
|
|
"present_in_text": false,
|
|
"tokens": [
|
|
"output",
|
|
"embedding"
|
|
],
|
|
"ocr_run": []
|
|
},
|
|
{
|
|
"label": "Masked Multi-Head Attention",
|
|
"ocr_status": "OCR_BOUNDARY",
|
|
"present_in_text": false,
|
|
"tokens": [
|
|
"masked",
|
|
"multi",
|
|
"head",
|
|
"attention"
|
|
],
|
|
"ocr_run": []
|
|
},
|
|
{
|
|
"label": "Linear",
|
|
"ocr_status": "OCR_EXACT",
|
|
"present_in_text": false,
|
|
"tokens": [
|
|
"linear"
|
|
],
|
|
"ocr_run": [
|
|
{
|
|
"text": "Linear",
|
|
"x": 964,
|
|
"y": 365
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"label": "Softmax",
|
|
"ocr_status": "OCR_EXACT",
|
|
"present_in_text": true,
|
|
"tokens": [
|
|
"softmax"
|
|
],
|
|
"ocr_run": [
|
|
{
|
|
"text": "l",
|
|
"x": 845,
|
|
"y": 215
|
|
},
|
|
{
|
|
"text": "|",
|
|
"x": 1198,
|
|
"y": 215
|
|
},
|
|
{
|
|
"text": "Softmax",
|
|
"x": 936,
|
|
"y": 224
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"label": "Output Probabilities",
|
|
"ocr_status": "OCR_BOUNDARY",
|
|
"present_in_text": false,
|
|
"tokens": [
|
|
"output",
|
|
"probabilities"
|
|
],
|
|
"ocr_run": []
|
|
}
|
|
],
|
|
"ocr_groups": [
|
|
[
|
|
{
|
|
"text": "Output",
|
|
"x": 939,
|
|
"y": 8
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Probabilities",
|
|
"x": 873,
|
|
"y": 83
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "t",
|
|
"x": 1017,
|
|
"y": 140
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "l",
|
|
"x": 845,
|
|
"y": 215
|
|
},
|
|
{
|
|
"text": "|",
|
|
"x": 1198,
|
|
"y": 215
|
|
},
|
|
{
|
|
"text": "Softmax",
|
|
"x": 936,
|
|
"y": 224
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "(",
|
|
"x": 845,
|
|
"y": 355
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "J",
|
|
"x": 1197,
|
|
"y": 355
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Linear",
|
|
"x": 964,
|
|
"y": 365
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "J",
|
|
"x": 1017,
|
|
"y": 433
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "|",
|
|
"x": 845,
|
|
"y": 511
|
|
},
|
|
{
|
|
"text": "}<s",
|
|
"x": 1199,
|
|
"y": 511
|
|
},
|
|
{
|
|
"text": "Add",
|
|
"x": 887,
|
|
"y": 521
|
|
},
|
|
{
|
|
"text": "&",
|
|
"x": 998,
|
|
"y": 521
|
|
},
|
|
{
|
|
"text": "Norm",
|
|
"x": 1047,
|
|
"y": 521
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Feed",
|
|
"x": 976,
|
|
"y": 614
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Forward",
|
|
"x": 939,
|
|
"y": 678
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "yy,",
|
|
"x": 1269,
|
|
"y": 770
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "|",
|
|
"x": 845,
|
|
"y": 825
|
|
},
|
|
{
|
|
"text": "eA",
|
|
"x": 1196,
|
|
"y": 825
|
|
},
|
|
{
|
|
"text": "Add",
|
|
"x": 887,
|
|
"y": 835
|
|
},
|
|
{
|
|
"text": "&",
|
|
"x": 998,
|
|
"y": 835
|
|
},
|
|
{
|
|
"text": "Norm",
|
|
"x": 1047,
|
|
"y": 835
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "|",
|
|
"x": 657,
|
|
"y": 882
|
|
},
|
|
{
|
|
"text": "Add",
|
|
"x": 347,
|
|
"y": 892
|
|
},
|
|
{
|
|
"text": "&Norm",
|
|
"x": 458,
|
|
"y": 892
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Multi-Head",
|
|
"x": 908,
|
|
"y": 928
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Feed",
|
|
"x": 436,
|
|
"y": 982
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Attention",
|
|
"x": 927,
|
|
"y": 992
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Forward",
|
|
"x": 399,
|
|
"y": 1045
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Nx",
|
|
"x": 1387,
|
|
"y": 1045
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "j}",
|
|
"x": 990,
|
|
"y": 1054
|
|
},
|
|
{
|
|
"text": "Sf",
|
|
"x": 1111,
|
|
"y": 1054
|
|
},
|
|
{
|
|
"text": "_}",
|
|
"x": 832,
|
|
"y": 1058
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "L",
|
|
"x": 477,
|
|
"y": 1127
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "|",
|
|
"x": 845,
|
|
"y": 1191
|
|
},
|
|
{
|
|
"text": "Jen",
|
|
"x": 1198,
|
|
"y": 1191
|
|
},
|
|
{
|
|
"text": "Add",
|
|
"x": 887,
|
|
"y": 1201
|
|
},
|
|
{
|
|
"text": "&",
|
|
"x": 998,
|
|
"y": 1201
|
|
},
|
|
{
|
|
"text": "Norm",
|
|
"x": 1047,
|
|
"y": 1201
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Nx",
|
|
"x": 47,
|
|
"y": 1234
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "|",
|
|
"x": 658,
|
|
"y": 1251
|
|
},
|
|
{
|
|
"text": "Add",
|
|
"x": 347,
|
|
"y": 1261
|
|
},
|
|
{
|
|
"text": "&Norm",
|
|
"x": 458,
|
|
"y": 1261
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Masked",
|
|
"x": 943,
|
|
"y": 1291
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Multi-Head",
|
|
"x": 369,
|
|
"y": 1350
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Multi-Head",
|
|
"x": 908,
|
|
"y": 1355
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Attention",
|
|
"x": 387,
|
|
"y": 1414
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Attention",
|
|
"x": 935,
|
|
"y": 1418
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "A__f",
|
|
"x": 345,
|
|
"y": 1483
|
|
},
|
|
{
|
|
"text": "=i",
|
|
"x": 884,
|
|
"y": 1483
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Vas:",
|
|
"x": 1000,
|
|
"y": 1573
|
|
},
|
|
{
|
|
"text": "\\",
|
|
"x": 179,
|
|
"y": 1581
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "N",
|
|
"x": 812,
|
|
"y": 1575
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Positional",
|
|
"x": 5,
|
|
"y": 1661
|
|
},
|
|
{
|
|
"text": "OF",
|
|
"x": 288,
|
|
"y": 1665
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Positional",
|
|
"x": 1269,
|
|
"y": 1672
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Encoding",
|
|
"x": 5,
|
|
"y": 1736
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Encoding",
|
|
"x": 1269,
|
|
"y": 1747
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Output",
|
|
"x": 952,
|
|
"y": 1834
|
|
},
|
|
{
|
|
"text": "Input",
|
|
"x": 436,
|
|
"y": 1835
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Embedding",
|
|
"x": 362,
|
|
"y": 1899
|
|
},
|
|
{
|
|
"text": "Embedding",
|
|
"x": 902,
|
|
"y": 1899
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "Outputs",
|
|
"x": 924,
|
|
"y": 2101
|
|
},
|
|
{
|
|
"text": "Inputs",
|
|
"x": 412,
|
|
"y": 2102
|
|
}
|
|
],
|
|
[
|
|
{
|
|
"text": "(shifted",
|
|
"x": 862,
|
|
"y": 2175
|
|
},
|
|
{
|
|
"text": "right)",
|
|
"x": 1071,
|
|
"y": 2175
|
|
}
|
|
]
|
|
]
|
|
}
|