igl_3states_1.stderr

creating quadratic features for pairs: UA
final_regressor = igl_1.vw
predictions = igl_1.pred
Enabling FTRL based optimization
Algorithm used: Coin Betting
ftrl_alpha = 4
ftrl_beta = 1
using no cache
Reading datafile = train-sets/igl_3states_1.dsjson
num sources = 1
Num weight bits = 18
learning rate = 0.5
initial_t = 0
power_t = 0.5
cb_type = mtr
Enabled learners: ftrl-Coin Betting, scorer-identity, csoaa_ldf-rank, cb_adf, cb_explore_adf_greedy, experimental_igl, shared_feature_merger
Input label = CB_WITH_OBSERVATIONS
Output pred = ACTION_PROBS
average  since         example        example        current        current  current
loss     last          counter         weight          label        predict features
0.000000 0.000000            1            1.0       3:0:0.25         0:0.25       21
0.000000 0.000000            2            2.0       1:0:0.25         0:0.25       21
0.000000 0.000000            4            4.0       2:0:0.25         0:0.25       21
-0.03676 -0.07352            8            8.0       2:0:0.85         0:0.25       21

finished run
number of examples = 10
weighted example sum = 10.000000
weighted label sum = 0.000000
average loss = -0.029412
total feature number = 280