{ "validation": { "overlap_argmax": { "precision": 0.6756756756756757, "recall": 0.6857142857142857, "f1": 0.6806579693703914, "number": 105, "threshold": null, "corrected_fp_per_1k": 0.4957411329938258, "_counts": { "overall": [ 105, 72, 111, 75 ], "GRAM": [ 51, 20, 50, 23 ], "SCRIPT": [ 54, 52, 61, 52 ], "_corrected": [ 33, 66567 ] }, "GRAM": { "precision": 0.46, "recall": 0.39215686274509803, "f1": 0.4233778186838472, "number": 51 }, "SCRIPT": { "precision": 0.8524590163934426, "recall": 0.9629629629629629, "f1": 0.9043478260869565, "number": 54 } }, "overlap_tuned": { "precision": 0.8714285714285714, "recall": 0.580952380952381, "f1": 0.6971428571428571, "number": 105, "threshold": [ 0.99, 0.98 ], "corrected_fp_per_1k": 0.10515721002899335, "_counts": { "overall": [ 105, 61, 70, 61 ], "GRAM": [ 51, 11, 19, 11 ], "SCRIPT": [ 54, 50, 51, 50 ], "_corrected": [ 7, 66567 ] }, "GRAM": { "precision": 0.5789473684210527, "recall": 0.21568627450980393, "f1": 0.3142857142857143, "number": 51 }, "SCRIPT": { "precision": 0.9803921568627451, "recall": 0.9259259259259259, "f1": 0.9523809523809523, "number": 54 } }, "seqeval_at_threshold": { "precision": 0.7428571428571429, "recall": 0.49523809523809526, "f1": 0.5942857142857143, "threshold": [ 0.99, 0.98 ], "corrected_fp_per_1k": 0.10515721002899335, "GRAM": { "precision": 0.5263157894736842, "recall": 0.19607843137254902, "f1": 0.2857142857142857, "number": 51.0 }, "SCRIPT": { "precision": 0.8235294117647058, "recall": 0.7777777777777778, "f1": 0.7999999999999999, "number": 54.0 } }, "events": { "threshold": [ 0.99, 0.98 ], "event_recall": 0.580952380952381, "span_precision": 0.9104477611940298, "event_f1": 0.7093023255813954, "head_recall": 0.5714285714285714, "events": 105, "pred_spans": 67, "row_precision": 0.9782608695652174, "row_recall": 0.7894736842105263, "rows_tp_fp_fn_tn": [ 45, 1, 12, 42 ], "corrected_fp_per_1k": 0.09013475145342287, "event_recall_tol2": 0.5904761904761905, "span_precision_tol2": 0.9253731343283582, "event_f1_tol2": 0.7209302325581395, "recall_by_type": { "AGREEMENT": { "events": 15, "recall": 0.4, "recall_tol2": 0.4 }, "CJK": { "events": 15, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 13, "recall": 0.15384615384615385, "recall_tol2": 0.15384615384615385 }, "LATIN_INSERT": { "events": 29, "recall": 0.896551724137931, "recall_tol2": 0.896551724137931 }, "MIXED_SCRIPT": { "events": 10, "recall": 0.9, "recall_tol2": 0.9 }, "MORPHOLOGY": { "events": 6, "recall": 0.0, "recall_tol2": 0.16666666666666666 }, "OMISSION": { "events": 6, "recall": 0.16666666666666666, "recall_tol2": 0.16666666666666666 }, "ORTHOGRAPHY": { "events": 11, "recall": 0.18181818181818182, "recall_tol2": 0.18181818181818182 } }, "ap_token": 0.6236939033626719, "at_0.5": { "threshold": [ 0.5, 0.5 ], "event_recall": 0.6952380952380952, "span_precision": 0.7184466019417476, "event_f1": 0.7066518411930146, "head_recall": 0.6761904761904762, "events": 105, "pred_spans": 103, "row_precision": 0.9090909090909091, "row_recall": 0.8771929824561403, "rows_tp_fp_fn_tn": [ 50, 5, 7, 38 ], "corrected_fp_per_1k": 0.43565129869154384, "event_recall_tol2": 0.7047619047619048, "span_precision_tol2": 0.7281553398058253, "event_f1_tol2": 0.7162676647092986 } } }, "test": { "overlap_argmax": { "precision": 0.6260869565217392, "recall": 0.5964912280701754, "f1": 0.6109308709757924, "number": 114, "threshold": null, "corrected_fp_per_1k": 0.4441632905770085, "_counts": { "overall": [ 114, 68, 115, 72 ], "SCRIPT": [ 52, 50, 53, 50 ], "GRAM": [ 62, 17, 62, 21 ], "_corrected": [ 33, 74297 ] }, "SCRIPT": { "precision": 0.9433962264150944, "recall": 0.9615384615384616, "f1": 0.9523809523809524, "number": 52 }, "GRAM": { "precision": 0.3387096774193548, "recall": 0.27419354838709675, "f1": 0.30305602716468594, "number": 62 } }, "overlap_at_val_threshold": { "precision": 0.7763157894736842, "recall": 0.49122807017543857, "f1": 0.6017118921872153, "number": 114, "threshold": [ 0.99, 0.98 ], "corrected_fp_per_1k": 0.14805443019233616, "_counts": { "overall": [ 114, 56, 76, 59 ], "SCRIPT": [ 52, 48, 51, 49 ], "GRAM": [ 62, 7, 25, 9 ], "_corrected": [ 11, 74297 ] }, "SCRIPT": { "precision": 0.9607843137254902, "recall": 0.9230769230769231, "f1": 0.9415532425940754, "number": 52 }, "GRAM": { "precision": 0.36, "recall": 0.11290322580645161, "f1": 0.1718963165075034, "number": 62 } }, "seqeval_at_val_threshold": { "precision": 0.6052631578947368, "recall": 0.40350877192982454, "f1": 0.4842105263157894, "threshold": [ 0.99, 0.98 ], "corrected_fp_per_1k": 0.14805443019233616, "GRAM": { "precision": 0.12, "recall": 0.04838709677419355, "f1": 0.0689655172413793, "number": 62.0 }, "SCRIPT": { "precision": 0.8431372549019608, "recall": 0.8269230769230769, "f1": 0.8349514563106797, "number": 52.0 } }, "events": { "threshold": [ 0.99, 0.98 ], "event_recall": 0.49122807017543857, "span_precision": 0.8142857142857143, "event_f1": 0.6127855634478786, "head_recall": 0.47368421052631576, "events": 114, "pred_spans": 70, "row_precision": 0.9333333333333333, "row_recall": 0.7368421052631579, "rows_tp_fp_fn_tn": [ 42, 3, 15, 40 ], "corrected_fp_per_1k": 0.13459493653848742, "event_recall_tol2": 0.5175438596491229, "span_precision_tol2": 0.8571428571428571, "event_f1_tol2": 0.6453965360072926, "recall_by_type": { "AGREEMENT": { "events": 12, "recall": 0.16666666666666666, "recall_tol2": 0.25 }, "CJK": { "events": 11, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 18, "recall": 0.05555555555555555, "recall_tol2": 0.1111111111111111 }, "LATIN_INSERT": { "events": 30, "recall": 0.9, "recall_tol2": 0.9 }, "MIXED_SCRIPT": { "events": 11, "recall": 0.9090909090909091, "recall_tol2": 0.9090909090909091 }, "MORPHOLOGY": { "events": 14, "recall": 0.07142857142857142, "recall_tol2": 0.14285714285714285 }, "OMISSION": { "events": 7, "recall": 0.2857142857142857, "recall_tol2": 0.2857142857142857 }, "ORTHOGRAPHY": { "events": 11, "recall": 0.18181818181818182, "recall_tol2": 0.18181818181818182 } }, "ap_token": 0.5023163735078685, "at_0.5": { "threshold": [ 0.5, 0.5 ], "event_recall": 0.6052631578947368, "span_precision": 0.69, "event_f1": 0.6448598130841122, "head_recall": 0.5877192982456141, "events": 114, "pred_spans": 100, "row_precision": 0.9074074074074074, "row_recall": 0.8596491228070176, "rows_tp_fp_fn_tn": [ 49, 5, 8, 38 ], "corrected_fp_per_1k": 0.36340632865391603, "event_recall_tol2": 0.6403508771929824, "span_precision_tol2": 0.73, "event_f1_tol2": 0.6822429906542056 } } }, "threshold": [ 0.99, 0.98 ], "reload_check": { "rows": 3, "max_abs_prob_diff": 0.0, "ok": true }, "history": [ { "step": 360, "eval_overlap_f1": 0.6350515463917525, "eval_overlap_precision": 0.56, "eval_overlap_recall": 0.7333333333333333, "eval_seqeval_f1": 0.5254901960784313, "eval_corrected_fp_per_1k": 0.8262352216563763, "eval_ap_token": 0.6384448209177483, "eval_overlap_f1_SCRIPT": 0.8779141104294479, "eval_overlap_f1_GRAM": 0.3964338781575037, "eval_synth_overlap_f1": 0.7032445833782472, "eval_runtime": 12.659429788589478, "epoch": 1.0, "recall_by_type": { "AGREEMENT": { "events": 15, "recall": 0.6666666666666666, "recall_tol2": 0.6666666666666666 }, "CJK": { "events": 15, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 13, "recall": 0.3076923076923077, "recall_tol2": 0.3076923076923077 }, "LATIN_INSERT": { "events": 29, "recall": 1.0, "recall_tol2": 1.0 }, "MIXED_SCRIPT": { "events": 10, "recall": 1.0, "recall_tol2": 1.0 }, "MORPHOLOGY": { "events": 6, "recall": 0.3333333333333333, "recall_tol2": 0.5 }, "OMISSION": { "events": 6, "recall": 0.5, "recall_tol2": 0.5 }, "ORTHOGRAPHY": { "events": 11, "recall": 0.36363636363636365, "recall_tol2": 0.36363636363636365 } }, "synthetic_recall_by_type": { "AGREEMENT": { "events": 19, "recall": 0.9473684210526315, "recall_tol2": 0.9473684210526315 }, "CJK": { "events": 25, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 32, "recall": 0.84375, "recall_tol2": 0.84375 }, "LATIN_INSERT": { "events": 68, "recall": 1.0, "recall_tol2": 1.0 }, "MIXED_SCRIPT": { "events": 30, "recall": 1.0, "recall_tol2": 1.0 }, "MORPHOLOGY": { "events": 25, "recall": 0.96, "recall_tol2": 0.96 }, "OMISSION": { "events": 15, "recall": 0.7333333333333333, "recall_tol2": 0.7333333333333333 }, "ORTHOGRAPHY": { "events": 20, "recall": 0.8, "recall_tol2": 0.8 }, "TRANSLITERATION": { "events": 6, "recall": 1.0, "recall_tol2": 1.0 } } }, { "step": 720, "eval_overlap_f1": 0.6462668298653611, "eval_overlap_precision": 0.6111111111111112, "eval_overlap_recall": 0.6857142857142857, "eval_seqeval_f1": 0.5627705627705628, "eval_corrected_fp_per_1k": 0.5558309672961077, "eval_ap_token": 0.6562105890304564, "eval_overlap_f1_SCRIPT": 0.8685786322092656, "eval_overlap_f1_GRAM": 0.3993344425956739, "eval_synth_overlap_f1": 0.7732217573221758, "eval_runtime": 11.265647411346436, "epoch": 2.0, "recall_by_type": { "AGREEMENT": { "events": 15, "recall": 0.6, "recall_tol2": 0.6 }, "CJK": { "events": 15, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 13, "recall": 0.3076923076923077, "recall_tol2": 0.3076923076923077 }, "LATIN_INSERT": { "events": 29, "recall": 0.9655172413793104, "recall_tol2": 0.9655172413793104 }, "MIXED_SCRIPT": { "events": 10, "recall": 0.9, "recall_tol2": 0.9 }, "MORPHOLOGY": { "events": 6, "recall": 0.0, "recall_tol2": 0.16666666666666666 }, "OMISSION": { "events": 6, "recall": 0.5, "recall_tol2": 0.5 }, "ORTHOGRAPHY": { "events": 11, "recall": 0.36363636363636365, "recall_tol2": 0.36363636363636365 } }, "synthetic_recall_by_type": { "AGREEMENT": { "events": 19, "recall": 0.8947368421052632, "recall_tol2": 0.9473684210526315 }, "CJK": { "events": 25, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 32, "recall": 0.84375, "recall_tol2": 0.84375 }, "LATIN_INSERT": { "events": 68, "recall": 1.0, "recall_tol2": 1.0 }, "MIXED_SCRIPT": { "events": 30, "recall": 1.0, "recall_tol2": 1.0 }, "MORPHOLOGY": { "events": 25, "recall": 0.96, "recall_tol2": 0.96 }, "OMISSION": { "events": 15, "recall": 0.8, "recall_tol2": 0.8 }, "ORTHOGRAPHY": { "events": 20, "recall": 0.8, "recall_tol2": 0.8 }, "TRANSLITERATION": { "events": 6, "recall": 1.0, "recall_tol2": 1.0 } } }, { "step": 1080, "eval_overlap_f1": 0.6752591455054022, "eval_overlap_precision": 0.6328125, "eval_overlap_recall": 0.7238095238095238, "eval_seqeval_f1": 0.575107296137339, "eval_corrected_fp_per_1k": 0.5708534258716782, "eval_ap_token": 0.6299657076548237, "eval_overlap_f1_SCRIPT": 0.8851090149992269, "eval_overlap_f1_GRAM": 0.42878560719640174, "eval_synth_overlap_f1": 0.7689594356261023, "eval_runtime": 12.065546751022339, "epoch": 3.0, "recall_by_type": { "AGREEMENT": { "events": 15, "recall": 0.6, "recall_tol2": 0.6 }, "CJK": { "events": 15, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 13, "recall": 0.38461538461538464, "recall_tol2": 0.38461538461538464 }, "LATIN_INSERT": { "events": 29, "recall": 1.0, "recall_tol2": 1.0 }, "MIXED_SCRIPT": { "events": 10, "recall": 1.0, "recall_tol2": 1.0 }, "MORPHOLOGY": { "events": 6, "recall": 0.0, "recall_tol2": 0.16666666666666666 }, "OMISSION": { "events": 6, "recall": 0.5, "recall_tol2": 0.5 }, "ORTHOGRAPHY": { "events": 11, "recall": 0.45454545454545453, "recall_tol2": 0.45454545454545453 } }, "synthetic_recall_by_type": { "AGREEMENT": { "events": 19, "recall": 0.8421052631578947, "recall_tol2": 0.8947368421052632 }, "CJK": { "events": 25, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 32, "recall": 0.78125, "recall_tol2": 0.8125 }, "LATIN_INSERT": { "events": 68, "recall": 1.0, "recall_tol2": 1.0 }, "MIXED_SCRIPT": { "events": 30, "recall": 1.0, "recall_tol2": 1.0 }, "MORPHOLOGY": { "events": 25, "recall": 0.96, "recall_tol2": 0.96 }, "OMISSION": { "events": 15, "recall": 0.6, "recall_tol2": 0.6 }, "ORTHOGRAPHY": { "events": 20, "recall": 0.75, "recall_tol2": 0.75 }, "TRANSLITERATION": { "events": 6, "recall": 1.0, "recall_tol2": 1.0 } } }, { "step": 1440, "eval_overlap_f1": 0.6806579693703914, "eval_overlap_precision": 0.6756756756756757, "eval_overlap_recall": 0.6857142857142857, "eval_seqeval_f1": 0.5833333333333333, "eval_corrected_fp_per_1k": 0.4957411329938258, "eval_ap_token": 0.6236939033626719, "eval_overlap_f1_SCRIPT": 0.9043478260869565, "eval_overlap_f1_GRAM": 0.4233778186838472, "eval_synth_overlap_f1": 0.7998049699604156, "eval_runtime": 11.419092655181885, "epoch": 4.0, "recall_by_type": { "AGREEMENT": { "events": 15, "recall": 0.5333333333333333, "recall_tol2": 0.5333333333333333 }, "CJK": { "events": 15, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 13, "recall": 0.38461538461538464, "recall_tol2": 0.38461538461538464 }, "LATIN_INSERT": { "events": 29, "recall": 1.0, "recall_tol2": 1.0 }, "MIXED_SCRIPT": { "events": 10, "recall": 0.9, "recall_tol2": 0.9 }, "MORPHOLOGY": { "events": 6, "recall": 0.0, "recall_tol2": 0.16666666666666666 }, "OMISSION": { "events": 6, "recall": 0.5, "recall_tol2": 0.5 }, "ORTHOGRAPHY": { "events": 11, "recall": 0.36363636363636365, "recall_tol2": 0.36363636363636365 } }, "synthetic_recall_by_type": { "AGREEMENT": { "events": 19, "recall": 0.8421052631578947, "recall_tol2": 0.8947368421052632 }, "CJK": { "events": 25, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 32, "recall": 0.78125, "recall_tol2": 0.84375 }, "LATIN_INSERT": { "events": 68, "recall": 1.0, "recall_tol2": 1.0 }, "MIXED_SCRIPT": { "events": 30, "recall": 1.0, "recall_tol2": 1.0 }, "MORPHOLOGY": { "events": 25, "recall": 0.96, "recall_tol2": 0.96 }, "OMISSION": { "events": 15, "recall": 0.6, "recall_tol2": 0.6 }, "ORTHOGRAPHY": { "events": 20, "recall": 0.75, "recall_tol2": 0.8 }, "TRANSLITERATION": { "events": 6, "recall": 1.0, "recall_tol2": 1.0 } } }, { "step": 1800, "eval_overlap_f1": 0.6789926681542875, "eval_overlap_precision": 0.6818181818181818, "eval_overlap_recall": 0.6761904761904762, "eval_seqeval_f1": 0.5953488372093023, "eval_corrected_fp_per_1k": 0.43565129869154384, "eval_ap_token": 0.6231596993104175, "eval_overlap_f1_SCRIPT": 0.907419150285352, "eval_overlap_f1_GRAM": 0.3981042654028436, "eval_synth_overlap_f1": 0.8043548911277217, "eval_runtime": 11.43203330039978, "epoch": 5.0, "recall_by_type": { "AGREEMENT": { "events": 15, "recall": 0.5333333333333333, "recall_tol2": 0.5333333333333333 }, "CJK": { "events": 15, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 13, "recall": 0.3076923076923077, "recall_tol2": 0.3076923076923077 }, "LATIN_INSERT": { "events": 29, "recall": 1.0, "recall_tol2": 1.0 }, "MIXED_SCRIPT": { "events": 10, "recall": 0.9, "recall_tol2": 0.9 }, "MORPHOLOGY": { "events": 6, "recall": 0.0, "recall_tol2": 0.16666666666666666 }, "OMISSION": { "events": 6, "recall": 0.5, "recall_tol2": 0.5 }, "ORTHOGRAPHY": { "events": 11, "recall": 0.2727272727272727, "recall_tol2": 0.2727272727272727 } }, "synthetic_recall_by_type": { "AGREEMENT": { "events": 19, "recall": 0.8421052631578947, "recall_tol2": 0.8947368421052632 }, "CJK": { "events": 25, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 32, "recall": 0.8125, "recall_tol2": 0.84375 }, "LATIN_INSERT": { "events": 68, "recall": 1.0, "recall_tol2": 1.0 }, "MIXED_SCRIPT": { "events": 30, "recall": 1.0, "recall_tol2": 1.0 }, "MORPHOLOGY": { "events": 25, "recall": 0.96, "recall_tol2": 0.96 }, "OMISSION": { "events": 15, "recall": 0.6666666666666666, "recall_tol2": 0.6666666666666666 }, "ORTHOGRAPHY": { "events": 20, "recall": 0.75, "recall_tol2": 0.8 }, "TRANSLITERATION": { "events": 6, "recall": 1.0, "recall_tol2": 1.0 } } }, { "step": 2160, "eval_overlap_f1": 0.675932977913176, "eval_overlap_precision": 0.6756756756756757, "eval_overlap_recall": 0.6761904761904762, "eval_seqeval_f1": 0.5925925925925927, "eval_corrected_fp_per_1k": 0.43565129869154384, "eval_ap_token": 0.6248177002304657, "eval_overlap_f1_SCRIPT": 0.899858512812451, "eval_overlap_f1_GRAM": 0.3981042654028436, "eval_synth_overlap_f1": 0.8189940436796824, "eval_runtime": 11.291409015655518, "epoch": 6.0, "recall_by_type": { "AGREEMENT": { "events": 15, "recall": 0.5333333333333333, "recall_tol2": 0.5333333333333333 }, "CJK": { "events": 15, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 13, "recall": 0.3076923076923077, "recall_tol2": 0.3076923076923077 }, "LATIN_INSERT": { "events": 29, "recall": 1.0, "recall_tol2": 1.0 }, "MIXED_SCRIPT": { "events": 10, "recall": 0.9, "recall_tol2": 0.9 }, "MORPHOLOGY": { "events": 6, "recall": 0.0, "recall_tol2": 0.16666666666666666 }, "OMISSION": { "events": 6, "recall": 0.5, "recall_tol2": 0.5 }, "ORTHOGRAPHY": { "events": 11, "recall": 0.2727272727272727, "recall_tol2": 0.2727272727272727 } }, "synthetic_recall_by_type": { "AGREEMENT": { "events": 19, "recall": 0.8421052631578947, "recall_tol2": 0.8947368421052632 }, "CJK": { "events": 25, "recall": 1.0, "recall_tol2": 1.0 }, "GOVERNMENT": { "events": 32, "recall": 0.8125, "recall_tol2": 0.8125 }, "LATIN_INSERT": { "events": 68, "recall": 1.0, "recall_tol2": 1.0 }, "MIXED_SCRIPT": { "events": 30, "recall": 1.0, "recall_tol2": 1.0 }, "MORPHOLOGY": { "events": 25, "recall": 0.96, "recall_tol2": 0.96 }, "OMISSION": { "events": 15, "recall": 0.6666666666666666, "recall_tol2": 0.6666666666666666 }, "ORTHOGRAPHY": { "events": 20, "recall": 0.75, "recall_tol2": 0.8 }, "TRANSLITERATION": { "events": 6, "recall": 1.0, "recall_tol2": 1.0 } } } ] }