Open uladkasach opened 7 years ago
Word vectors trained on 5.6M words did not perform well. Google's trained on 3 billion and perform well. They chose the top 3 million words. The frequency of the top 3million is likely quite large.
Word vectors trained on 5.6M words did not perform well. Google's trained on 3 billion and perform well. They chose the top 3 million words. The frequency of the top 3million is likely quite large.