language model 0.00322108
backoff model 0.002840483
first word 0.002337759
trigram model 0.002267349
word tokens 0.0022385829999999997
language models 0.0022090119999999998
word list 0.002166547
guage model 0.002153485
tbo model 0.00214135
model fit 0.002137266
word ids 0.002090414
word types 0.002082183
wrong word 0.0020804969999999997
model 0.00188613
language modeling 0.0018308369999999999
training data 0.00176242
contextual language 0.0017312599999999999
trigram language 0.001716169
statistical language 0.0016880909999999999
small language 0.001680427
data structure 0.00163842
hash function 0.0016270289999999999
much data 0.001592117
same memory 0.001591738
other words 0.001565825
straightforward data 0.0014980129999999999
standard backoff 0.001393425
same value 0.0013821789999999999
single hash 0.001358106
language 0.00133495
hash table 0.001328625
backoff weights 0.001302706
backoff estimate 0.001292551
large corpus 0.001291918
count probabilities 0.001284644
katz backoff 0.001278182
little memory 0.001268914
hash codes 0.001266446
hash range 0.0012593980000000001
stolcke pruning 0.001238627
backoff weight 0.001231524
bigram array 0.001228483
little pruning 0.001222876
much pruning 0.001220918
fixed memory 0.001216969
other compression 0.0012129480000000002
backoff alphas 0.00121283
pruning losses 0.001197056
subsequent hash 0.001188339
enough memory 0.001186531
memory sizes 0.0011840689999999998
memory allocation 0.0011810079999999999
memory hogs 0.0011794079999999998
hash collisions 0.001176857
memory constraint 0.001176576
memory budget 0.00117545
same number 0.001173525
unlimited memory 0.001171493
pruning criterion 0.001135053
other lan 0.001108394
loss function 0.001108263
other concerns 0.001102416
other implementa 0.0010956170000000001
other coeffi 0.001088347
bigram node 0.0010809650000000001
large set 0.001077471
input method 0.0010750870000000002
second array 0.001071023
same way 0.001062743
figure array 0.0010613950000000001
hashtbo method 0.0010573190000000001
same estimates 0.001054781
discounted probability 0.001053376
unigram array 0.001052865
probability mass 0.001049559
large integer 0.001037811
coding method 0.001033325
boundary value 0.001025064
test process 0.001004385
lion words 0.0010023550000000001
likelihood values 0.001001192
bigram nodes 0.001000906
trigram array 9.98631E-4
method editor 9.96244E-4
gram probabilities 9.805389999999999E-4
small counts 9.79536E-4
bigram case 9.70728E-4
pected value 9.61197E-4
same amount 9.57492E-4
address space 9.57491E-4
baseline system 9.55378E-4
backoff 9.54353E-4
same val 9.53799E-4
vocabulary size 9.490540000000001E-4
test set 9.45188E-4
ziptbo values 9.448600000000001E-4
exponential probabilities 9.44326E-4
discounted probabilities 9.32315E-4
training set 9.29841E-4
hashtbo values 9.26944E-4
