📄 trainlog.txt_second-ordercrfs
字号:
OPTION VALUES:Model directory: ./Training data file: train.taggedTesting data file: test.taggedUnlabeled data file: data.untaggedLabel representation: IOB2Model file: model.txtTraining log file (this one): trainlog.txtSecond-order Markov CRFsNumber of labels: 4Number of training sequences: 412Number of testing sequences: 86Number of unlabeled sequences: 0Number of context predicates: 17487Number of features: 30283Feature rare threshold: 1Context predicate rare threshold: 1Using multiple rare thresholds for features: 0Highlight feature: 0Number of training iterations: 60Initial lambda value: 0.0000Sigma square (for smoothing): 100.0000Epsilon for L-BFGS convergence: 0.000100Number of approximated hessian matrixes: 7Start to train ...Iteration: 1 Log-likelihood = -26641.805032 Norm(log-likelihood gradient vector) = 5547.125079 Norm(lambda vector) = 0.000000 Iteration elapsed: 3 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- b-np 529 486 368 75.72 69.57 72.51 o 725 829 646 77.93 89.10 83.14 i-np 691 630 518 82.22 74.96 78.43 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.62 77.88 78.25 Avg2. 1945 1945 1532 78.77 78.77 78.77 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 529 486 277 57.00 52.36 54.58 ----- ------ ----- ----- ------- ------- ------------- Avg1. 57.00 52.36 54.58 Avg2. 529 486 277 57.00 52.36 54.58 Current max chunk-based F1: 54.58 (iteration 1) Training iteration elapsed (including testing & evaluation time): 3 secondsIteration: 2 Log-likelihood = -21394.504556 Norm(log-likelihood gradient vector) = 4828.454251 Norm(lambda vector) = 1.000000 Iteration elapsed: 3 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- b-np 529 479 391 81.63 73.91 77.58 o 725 799 660 82.60 91.03 86.61 i-np 691 667 559 83.81 80.90 82.33 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.68 81.95 82.31 Avg2. 1945 1945 1610 82.78 82.78 82.78 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 529 479 303 63.26 57.28 60.12 ----- ------ ----- ----- ------- ------- ------------- Avg1. 63.26 57.28 60.12 Avg2. 529 479 303 63.26 57.28 60.12 Current max chunk-based F1: 60.12 (iteration 2) Training iteration elapsed (including testing & evaluation time): 3 secondsIteration: 3 Log-likelihood = -8730.879901 Norm(log-likelihood gradient vector) = 3259.885589 Norm(lambda vector) = 9.709716 Iteration elapsed: 3 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- b-np 529 408 382 93.63 72.21 81.54 o 725 800 693 86.62 95.59 90.89 i-np 691 737 631 85.62 91.32 88.38 ----- ------ ----- ----- ------- ------- ------------- Avg1. 88.62 86.37 87.48 Avg2. 1945 1945 1706 87.71 87.71 87.71 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 529 408 310 75.98 58.60 66.17 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.98 58.60 66.17 Avg2. 529 408 310 75.98 58.60 66.17 Current max chunk-based F1: 66.17 (iteration 3) Training iteration elapsed (including testing & evaluation time): 3 secondsIteration: 4 Log-likelihood = -6525.602824 Norm(log-likelihood gradient vector) = 1575.238232 Norm(lambda vector) = 9.457559 Iteration elapsed: 3 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- b-np 529 501 457 91.22 86.39 88.74 o 725 776 708 91.24 97.66 94.34 i-np 691 668 611 91.47 88.42 89.92 ----- ------ ----- ----- ------- ------- ------------- Avg1. 91.31 90.82 91.06 Avg2. 1945 1945 1776 91.31 91.31 91.31 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 529 501 384 76.65 72.59 74.56 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.65 72.59 74.56 Avg2. 529 501 384 76.65 72.59 74.56 Current max chunk-based F1: 74.56 (iteration 4) Training iteration elapsed (including testing & evaluation time): 3 secondsIteration: 5 Log-likelihood = -6046.275745 Norm(log-likelihood gradient vector) = 996.018392 Norm(lambda vector) = 9.011274 Iteration elapsed: 3 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- b-np 529 516 471 91.28 89.04 90.14 o 725 795 714 89.81 98.48 93.95 i-np 691 634 598 94.32 86.54 90.26 ----- ------ ----- ----- ------- ------- ------------- Avg1. 91.80 91.35 91.58 Avg2. 1945 1945 1783 91.67 91.67 91.67 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 529 516 410 79.46 77.50 78.47 ----- ------ ----- ----- ------- ------- ------------- Avg1. 79.46 77.50 78.47 Avg2. 529 516 410 79.46 77.50 78.47 Current max chunk-based F1: 78.47 (iteration 5) Training iteration elapsed (including testing & evaluation time): 3 secondsIteration: 6 Log-likelihood = -5707.796577 Norm(log-likelihood gradient vector) = 680.797163 Norm(lambda vector) = 9.068814 Iteration elapsed: 3 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- b-np 529 527 491 93.17 92.82 92.99 o 725 774 717 92.64 98.90 95.66 i-np 691 644 622 96.58 90.01 93.18 ----- ------ ----- ----- ------- ------- ------------- Avg1. 94.13 93.91 94.02 Avg2. 1945 1945 1830 94.09 94.09 94.09 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 529 527 452 85.77 85.44 85.61 ----- ------ ----- ----- ------- ------- ------------- Avg1. 85.77 85.44 85.61 Avg2. 529 527 452 85.77 85.44 85.61 Current max chunk-based F1: 85.61 (iteration 6) Training iteration elapsed (including testing & evaluation time): 3 secondsIteration: 7 Log-likelihood = -5162.021610 Norm(log-likelihood gradient vector) = 665.946239 Norm(lambda vector) = 10.028874 Iteration elapsed: 3 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- b-np 529 524 494 94.27 93.38 93.83 o 725 774 721 93.15 99.45 96.20 i-np 691 647 630 97.37 91.17 94.17 ----- ------ ----- ----- ------- ------- ------------- Avg1. 94.93 94.67 94.80 Avg2. 1945 1945 1845 94.86 94.86 94.86 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 529 524 459 87.60 86.77 87.18 ----- ------ ----- ----- ------- ------- ------------- Avg1. 87.60 86.77 87.18 Avg2. 529 524 459 87.60 86.77 87.18 Current max chunk-based F1: 87.18 (iteration 7) Training iteration elapsed (including testing & evaluation time): 3 secondsIteration: 8 Log-likelihood = -4480.794556 Norm(log-likelihood gradient vector) = 516.549106 Norm(lambda vector) = 11.715696 Iteration elapsed: 3 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- b-np 529 507 492 97.04 93.01 94.98 o 725 734 708 96.46 97.66 97.05 i-np 691 704 664 94.32 96.09 95.20 ----- ------ ----- ----- ------- ------- ------------- Avg1. 95.94 95.58 95.76 Avg2. 1945 1945 1864 95.84 95.84 95.84 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 529 507 457 90.14 86.39 88.22 ----- ------ ----- ----- ------- ------- ------------- Avg1. 90.14 86.39 88.22 Avg2. 529 507 457 90.14 86.39 88.22 Current max chunk-based F1: 88.22 (iteration 8) Training iteration elapsed (including testing & evaluation time): 3 secondsIteration: 9 Log-likelihood = -3680.402424 Norm(log-likelihood gradient vector) = 742.650631 Norm(lambda vector) = 15.933625 Iteration elapsed: 3 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- b-np 529 525 504 96.00 95.27 95.64 o 725 759 722 95.13 99.59 97.30 i-np 691 661 646 97.73 93.49 95.56 ----- ------ ----- ----- ------- ------- ------------- Avg1. 96.29 96.12 96.20 Avg2. 1945 1945 1872 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 529 525 475 90.48 89.79 90.13 ----- ------ ----- ----- ------- ------- ------------- Avg1. 90.48 89.79 90.13 Avg2. 529 525 475 90.48 89.79 90.13 Current max chunk-based F1: 90.13 (iteration 9) Training iteration elapsed (including testing & evaluation time): 3 secondsIteration: 10 Log-likelihood = -3304.002286 Norm(log-likelihood gradient vector) = 295.224214 Norm(lambda vector) = 17.343823 Iteration elapsed: 3 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- b-np 529 529 503 95.09 95.09 95.09 o 725 765 723 94.51 99.72 97.05 i-np 691 651 638 98.00 92.33 95.08 ----- ------ ----- ----- ------- ------- ------------- Avg1. 95.87 95.71 95.79 Avg2. 1945 1945 1864 95.84 95.84 95.84 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 529 529 473 89.41 89.41 89.41 ----- ------ ----- ----- ------- ------- ------------- Avg1. 89.41 89.41 89.41 Avg2. 529 529 473 89.41 89.41 89.41 Current max chunk-based F1: 90.13 (iteration 9) Training iteration elapsed (including testing & evaluation time): 3 secondsIteration: 11 Log-likelihood = -3081.693064 Norm(log-likelihood gradient vector) = 321.707885 Norm(lambda vector) = 19.516581 Iteration elapsed: 3 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- b-np 529 532 505 94.92 95.46 95.19 o 725 766 723 94.39 99.72 96.98 i-np 691 647 637 98.45 92.19 95.22 ----- ------ ----- ----- ------- ------- ------------- Avg1. 95.92 95.79 95.86 Avg2. 1945 1945 1865 95.89 95.89 95.89 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 529 532 477 89.66 90.17 89.92 ----- ------ ----- ----- ------- ------- ------------- Avg1. 89.66 90.17 89.92 Avg2. 529 532 477 89.66 90.17 89.92 Current max chunk-based F1: 90.13 (iteration 9) Training iteration elapsed (including testing & evaluation time): 3 secondsIteration: 12 Log-likelihood = -2811.846856 Norm(log-likelihood gradient vector) = 292.588814 Norm(lambda vector) = 22.361233 Iteration elapsed: 3 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%)⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -