4 ms·
Maybe something like K-BERT, the idea is simple enough, and can leverage pretrained transformer models. https://arxiv.org/abs/1909.07606 https://arxiv.org/abs/
by phreeza 7y ago
Maybe something like K-BERT, the idea is simple enough, and can leverage pretrained transformer models.
https://arxiv.org/abs/1909.07606 https://arxiv.org/abs/1909.07606