Взвешенное комбинирование грамм-матриц

AbstractMulti-modal models that learn semantic representations from both linguistic and perceptual input outperform language-only models on a range of evaluations, and better reflect human concept acquisition. Most perceptual input to …...