distribution model 0.00315379
language model 0.003109362
topic model 0.003011542
same word 0.0027555770000000004
selection model 0.0026029039999999996
probabilistic model 0.0025853679999999998
generation model 0.002565992
model ter 0.002559681
annotation model 0.0025375009999999997
trigram model 0.002511203
model parameters 0.002500839
similarity word 0.0024928140000000003
model our 0.002485809
bigram model 0.0024810659999999997
word error 0.0024805
abstractive model 0.002474602
above model 0.0024665269999999996
general model 0.002466243
word distributions 0.002457517
model grammaticality 0.002453172
model estimates 0.0024485519999999997
guage model 0.002435064
tic model 0.002431839
notation model 0.0024182419999999997
inal model 0.0024182419999999997
channel model 0.0024182419999999997
word order 0.002384291
multimodal word 0.0023796200000000003
word overlap 0.0023572750000000003
word salad 0.002355386
outperform word 0.002355386
model 0.00220224
topic models 0.001904872
language sentence 0.001804531
topic distribution 0.001760852
document sentence 0.001713113
sentence topic 0.001706711
other words 0.001696427
other models 0.0016744669999999998
content words 0.001637283
news image 0.001628876
many words 0.00162238
image topic 0.0016012320000000002
function words 0.001530577
visual words 0.001524886
probabilistic models 0.0014786979999999999
generation models 0.0014593219999999999
description words 0.001457662
probability distribution 0.00145353
length distribution 0.001453347
same topic 0.001426289
sequence models 0.0014258839999999999
image caption 0.00142514
extractive models 0.00142176
words blue 0.001403828
caption data 0.001402637
sentence length 0.001399206
textual words 0.001398918
present models 0.001398266
single words 0.001383777
preceding words 0.001371751
abstractive models 0.0013679319999999999
abstract words 0.001353395
eration models 0.0013477279999999999
able models 0.001346209
ent models 0.0013391269999999999
similar sentence 0.001326029
document sentences 0.001324792
first sentence 0.00132452
document length 0.001317501
ation models 0.0013120319999999999
training data 0.0013108360000000001
image features 0.00130765
image captions 0.00125282
language description 0.001247254
first document 0.001242815
natural language 0.001217653
same corpus 0.001213167
average sentence 0.001209275
visual information 0.001200297
sentence baseline 0.001194151
same system 0.00119007
document phrases 0.001178924
probabilistic image 0.001175058
ability distribution 0.001170618
distribution pdmix 0.001169151
single sentence 0.0011636560000000001
news story 0.001154495
relevant sentence 0.001153909
news database 0.0011537399999999999
language processing 0.001153542
lead sentence 0.0011470620000000001
many caption 0.0011380600000000002
sentence extraction 0.001137994
same content 0.00113674
caption length 0.001135007
image structure 0.001133198
language descrip 0.001132113
image description 0.001132062
new data 0.001129688
