only testing
Num weight bits = 18
learning rate = 10
initial_t = 1
power_t = 0.5
predictions = bs.prreg.predict
using no cache
Reading datafile = train-sets/0002.dat
num sources = 1
average    since         example     example  current  current  current
loss       last          counter      weight    label  predict features
0.000319   0.000319            1         1.0   0.5211   0.5033       15
0.002416   0.004513            2         2.0   0.5353   0.4681       15
0.006433   0.010451            4         4.0   0.5854   0.4555       15
0.026282   0.046130            8         8.0   0.5575   0.4545       15
0.029274   0.032266           16        16.0   0.5878   0.4954       15
0.030773   0.032272           32        32.0   0.6038   0.4671       15
0.021776   0.012780           64        64.0   0.5683   0.5623       15
0.020985   0.020193          128       128.0   0.5351   0.5296       15
0.018010   0.015035          256       256.0   0.5385   0.4933       15
0.014488   0.010965          512       512.0   0.5053   0.4518       15
0.013376   0.012264         1024      1024.0   0.5750   0.5094       15
0.012045   0.010714         2048      2048.0   0.5204   0.5061       15
0.010188   0.008331         4096      4096.0   0.5042   0.4592       15
0.009317   0.008447         8192      8192.0   0.4967   0.4613       15
0.008218   0.007119        16384     16384.0   0.5011   0.5464       15
0.009231   0.010244        32768     32768.0   0.3915   0.5828       15
0.011318   0.013404        65536     65536.0   0.5043   0.4459       15

finished run
number of examples per pass = 74746
passes used = 1
weighted example sum = 69521
weighted label sum = 0
average loss = 0.0108557
best constant = -1.43843e-05
total feature number = 1119986