search

thunlp / SE-WRL

Improved Word Representation Learning with Sememes
MIT License
195 stars 56 forks source link

VocabFile problem #22

Closed bestpredicts closed 4 years ago

bestpredicts commented 5 years ago

Hi, when I add a word like '蛤蛤 50' to VocabFile , then I train model , but I get a error

dy@ubun:~/SE-WRL-master/datasets$ ./word2vec -train Clean-Sogou-ALL.txt -output vectors.bin -cbow 0 -size 200 -window 8 -negative 25 -hs 0 -sample 1e-4 -thre ads 30 -binary 1 -iter 3 -read-vocab VocabFile -read-meaning SememeFile -read-sense Word_Sense_Sememe_File -min-count 50 -alpha 0.025 Starting training using file Clean-Sogou-ALL.txt Vocab size: 462667 Words in train file: 2655924607 462667 1983 Alpha: 0.024999 Progress: 0.00% Words/thread/sec: 1.83k Segmentation fault how can I add words to VocabFile ?

heyLinsir commented 4 years ago

If you add new words to VocabFile, these words should be also added to Word_Sense_Sememe_File.