Evaluation tool for word embeddings
Wordsimilarity task is commonly used as a evaluation method for word-embeddings such as Word2vec, Glove. This task investigate how your vector capture semantics between word pairs.
Wordsim task can be run on english, spanish, and french wordembeddings.
python wordsim.py -l en -v vector_file.txt
Your vector file need to satisfy the following form.
- apple 0.01 0.2 0.01 ...
You can also feed in compressed version of vector file.
python wordsim.py -l en -v vector_file.txt.gz
- numpy, scipy, ( gzip, prettytable )