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
0.000331   0.000331            1         1.0   0.5211   0.5030       15
0.002435   0.004538            2         2.0   0.5353   0.4679       15
0.006879   0.011324            4         4.0   0.5854   0.4496       15
0.027055   0.047230            8         8.0   0.5575   0.4540       15
0.029899   0.032743           16        16.0   0.5878   0.4943       15
0.031465   0.033032           32        32.0   0.6038   0.4657       15
0.022291   0.013117           64        64.0   0.5683   0.5623       15
0.021452   0.020612          128       128.0   0.5351   0.5293       15
0.018406   0.015360          256       256.0   0.5385   0.4922       15
0.014747   0.011088          512       512.0   0.5053   0.4511       15
0.013612   0.012477         1024      1024.0   0.5750   0.5087       15
0.012210   0.010808         2048      2048.0   0.5204   0.5062       15
0.010295   0.008380         4096      4096.0   0.5042   0.4586       15
0.009424   0.008553         8192      8192.0   0.4967   0.4608       15
0.008262   0.007101        16384     16384.0   0.5011   0.5461       15
0.009269   0.010275        32768     32768.0   0.3915   0.5825       15
0.011352   0.013435        65536     65536.0   0.5043   0.4461       15

finished run
number of examples per pass = 74746
passes used = 1
weighted example sum = 69521
weighted label sum = 0
average loss = 0.010892
best constant = -1.43843e-05
total feature number = 1119986