word2vec_mt

word2vec embeddings for the Maltese language saved as a NumPy file. See https://github.com/mtanti/word2vec_mt for code that produced it. File descriptions: File name Description vocab_mt.txt The vocabulary of 249,080 tokens embedded as a line-separated text file. word2vec_mt.npy The Maltese word2vec embeddings. It is a NumPy file (open in Python with np.load) containing a matrix that has as many rows as the vocabulary size and 300 columns. The rows are the token embeddings of the token in the corresponding vocab_mt.txt line. word2vec_mten.npy The Maltese word2vec embeddings aligned to the English Google News 300 word2vec embeddings such that it should be possible to find the most similar English tokens to a given Maltese token. Does not work well. synonyms_mt.jsonl A human-readable small data set of Maltese synonyms (created manually by one person) JSON Lines file. synonyms_mt.json synonyms_mt.jsonl but split into val/dev/test data splits in a more machine-friendly format as used to evaluate the word2vec embeddings. The format is two parallel lists of source token and corresponding list of similar tokens. translations_mten.jsonl A human-readable small data set of Maltese-English word translations (created manually by one person) JSON Lines file. translations_mten.json translations_mten.jsonl but split into train/val/dev/test data splits in a more machine-friendly format as used to train and evaluate the aligned embeddings. The format is two parallel lists of source token and corresponding list of similar tokens.