Num weight bits = 18
learning rate = 10
initial_t = 1
power_t = 0.5
creating cache_file = train-sets/wsj_small.dat.gz.cache
Reading from train-sets/wsj_small.dat.gz
num sources = 1
average    since      sequence         example            current label      current predicted  current   cur   cur         predic.        examples
loss       last        counter          weight          sequence prefix        sequence prefix features  pass   pol            made          gener.
0.810811   0.810811          1       37.000000   [  1   2   3   1   4 ] [  1   1   1   1   1 ]     1654     0     0              37               0
0.750000   0.666667          2       64.000000   [ 11   2   3  11  11 ] [  9  11   9  11   9 ]     1194     0     0             841              37
0.709677   0.620690          3       93.000000   [ 14  10  13   9   1 ] [ 11  15  11   1   9 ]     1286     0     0            1465              64
0.713178   0.722222          4      129.000000   [  3   4   6   3   1 ] [ 11  11   2   3  11 ]     1608     0     0            2098              93
0.693750   0.612903          5      160.000000   [ 19   3  10   2   1 ] [  2   3   1   2   1 ]     1378     0     0            2903             129
0.658163   0.500000          6      196.000000   [ 19   2  22   4   3 ] [ 19   2  11  11  11 ]     1608     0     0            3619             160
0.659574   0.666667          7      235.000000   [ 10   2   3   1  10 ] [ 19   2  11   1   1 ]     1746     0     0            4428             196
0.594901   0.466102         12      353.000000   [  5  12  11  11  21 ] [ 11  12  29  21  21 ]     1102     0     0            7509             328
0.481534   0.367521         25      704.000000   [ 10  13  22   4   9 ] [ 10  13   1   4   3 ]     1148     0     0           15642             678
0.399859   0.319328         57     1418.000000   [ 19   1   4   6  36 ] [ 19   5   4   6   5 ]     2252     0     0           31220            1368

finished run
number of examples = 78
weighted example sum = 1932
weighted label sum = 0
average loss = 0.369
best constant = -0.0005179
total feature number = 85128