cbe_adf_first.stderr

predictions = cbe_adf_first.predict
Num weight bits = 18
learning rate = 0.5
initial_t = 0
power_t = 0.5
using no cache
Reading datafile = train-sets/cb_test.ldf
num sources = 1
Enabled reductions: gd, scorer, csoaa_ldf, cb_adf, cb_explore_adf_first, shared_feature_merger
average  since         example        example  current  current  current
loss     last          counter         weight    label  predict features
0.666667 0.666667            1            1.0  0:1:0.5        0:0.333333...        9
0.333333 0.000000            2            2.0  1:0:0.5        1:0.5...        6

finished run
number of examples = 3
weighted example sum = 3.000000
weighted label sum = 0.000000
average loss = 0.333333
total feature number = 23