feature set 0.002869767
language model 0.002720255
feature function 0.002550451
standard model 0.002523282
feature count 0.002382593
task model 0.002324918
feature constraints 0.002305115
feature weight 0.002301053
feature sets 0.002280325
arbitrary feature 0.002239369
ith feature 0.002213274
feature sharing 0.002201397
melm model 0.00217497
trigram model 0.002174621
same features 0.0021175129999999997
model dev 0.002099151
model ppl 0.002082419
training data 0.0020786019999999997
model interpolated 0.0020767249999999998
model interpolate 0.002074341
binarized model 0.002074341
training set 0.002067197
training corpus 0.002006257
feature 0.00197328
general features 0.0019311989999999998
trigram features 0.0019176109999999998
trigger features 0.0018968
model 0.00184678
complicated features 0.0018184179999999998
word history 0.0018003020000000002
word representation 0.001787676
next word 0.001727786
word error 0.001722088
word trigger 0.00171078
particular word 0.001695148
ith word 0.001643744
word represen 0.001631807
word trainer 0.001631807
features 0.00158977
training tokens 0.001579623
training examples 0.0015789419999999998
standard language 0.0015499770000000001
negative training 0.001544339
nnlm training 0.001520022
melm training 0.0014989
training melm 0.0014989
training cases 0.001469163
training instances 0.0014333789999999998
binary set 0.0014333649999999998
subsampled training 0.0014318949999999999
training patterns 0.00142827
test set 0.00142006
training exam 0.001404139
training instance 0.001398746
training takeso 0.001397758
language models 0.0013923529999999998
words history 0.0013734189999999999
set log 0.0013034029999999999
frequent words 0.001275807
text corpus 0.001257209
entropy language 0.001256881
data likelihood 0.001231284
language modeling 0.001217133
data sets 0.001214937
natural language 0.001213289
conditional probability 0.00120497
network language 0.001202959
standard models 0.00119538
language processing 0.001183261
ing corpus 0.001181608
sample data 0.001173874
training 0.00117071
subsampled data 0.001169077
validation data 0.001166805
validation set 0.0011554
enough data 0.001152276
treebank corpus 0.001145684
dation set 0.001131155
language mod 0.0011279419999999998
idation set 0.001123563
journal corpus 0.001108115
language modelers 0.001100714
approximating language 0.001100714
computational dependency 0.001098347
much probability 0.001095846
following probability 0.001085647
semantic information 0.001061666
standard subsampling 0.001051029
same results 0.001032939
large text 0.00102937
same learning 0.0010260970000000001
standard nnlm 0.0010258140000000001
probability definition 0.001022352
probability mass 0.001015618
probability defini 0.0010097349999999999
standard melm 0.0010046920000000002
different way 9.93691E-4
words 9.76867E-4
performance gain 9.75711E-4
large number 9.70376E-4
