Skip to content

Latest commit

 

History

History
 
 

test

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
data
data
 
 
en-de
en-de
 
 
en-ro
en-ro
 
 
load
load
 
 
.gitignore
.gitignore
 
 
README.md
README.md
 
 
test_score.py
test_score.py
 
 
test_train.sh
test_train.sh
 
 
test_train_bigmem.sh
test_train_bigmem.sh
 
 
test_train_domaininterpolation.sh
test_train_domaininterpolation.sh
 
 
test_train_drt.sh
test_train_drt.sh
 
 
test_train_drt_enc+decbase+dechigh_layernorm_weightnorm.sh
test_train_drt_enc+decbase+dechigh_layernorm_weightnorm.sh
 
 
test_train_drt_layernorm.sh
test_train_drt_layernorm.sh
 
 
test_train_drt_weightnorm.sh
test_train_drt_weightnorm.sh
 
 
test_train_float16.sh
test_train_float16.sh
 
 
test_train_float16_largebatch.sh
test_train_float16_largebatch.sh
 
 
test_train_float16_largebatch_largemodel.sh
test_train_float16_largebatch_largemodel.sh
 
 
test_train_float16_midbatch_largemodel.sh
test_train_float16_midbatch_largemodel.sh
 
 
test_train_largebatch.sh
test_train_largebatch.sh
 
 
test_train_largebatch_largemodel.sh
test_train_largebatch_largemodel.sh
 
 
test_train_layernorm.sh
test_train_layernorm.sh
 
 
test_train_midbatch_largemodel.sh
test_train_midbatch_largemodel.sh
 
 
test_train_verybigmem.sh
test_train_verybigmem.sh
 
 
test_train_weightnorm.sh
test_train_weightnorm.sh
 
 
test_translate.py
test_translate.py
 
 

README.md

Testing Nematus

To test translation, execute

THEANO_FLAGS=mode=FAST_RUN,floatX=float32,device=cpu python test_translate.py

To test scoring, execute

THEANO_FLAGS=mode=FAST_RUN,floatX=float32,device=cpu python test_score.py

more sample models (including scripts for pre- and postprocessing) are provided at: http://statmt.org/rsennrich/wmt16_systems/

to test training, execute

THEANO_FLAGS=mode=FAST_RUN,floatX=float32,device=cpu ./test_train.sh

note that the training script is just a toy setup to make sure the scripts run, and to allow for speed comparisons. For instructions to train a real-scale system, check the instructions at https://github.com/rsennrich/wmt16-scripts