creating quadratic features for pairs: Tf ff 
final_regressor = models/0002a.model
Num weight bits = 18
learning rate = 10
initial_t = 1
power_t = 0.5
using no cache
Reading from train-sets/0002.dat
num sources = 1
average    since         example     example  current  current  current
loss       last          counter      weight    label  predict features
0.142846   0.142846            3         3.0   0.5498   0.2156      197
0.090009   0.037171            6         6.0   0.2681   0.3538      197
0.055547   0.014192           11        11.0   0.4315   0.3707      197
0.054939   0.054331           22        22.0   0.5519   0.5857      197
0.030289   0.005640           44        44.0   0.5514   0.6277      197
0.022904   0.015346           87        87.0   0.5140   0.4772      197
0.015973   0.009043          174       174.0   0.5596   0.5397      197
0.011176   0.006380          348       348.0   0.5475   0.5475      197
0.007019   0.002861          696       696.0   0.3421   0.3701      197
0.004071   0.001124         1392      1392.0   0.4996   0.5633      197
0.002379   0.000687         2784      2784.0   0.5090   0.5174      197
0.001401   0.000422         5568      5568.0   0.6413   0.6156      197
0.000995   0.000588        11135     11135.0   0.3869   0.3979      197
0.000803   0.000611        22269     22269.0   0.5063   0.5071      197
0.000821   0.000839        44537     44537.0   0.4905   0.4853      197

finished run
number of examples = 74746
weighted example sum = 6.952e+04
weighted label sum = 3.511e+04
average loss = 0.0008412
best constant = 0.5051
best constant's loss = 0.25
total feature number = 14692454