3 ms·
An English pretrained model is available. Also, the SQuAD2.0 EM numbers are higher than the BERT results published in the BERT paper. Subsequent work has improv
by roseway4 7y ago
An English pretrained model is available. Also, the SQuAD2.0 EM numbers are higher than the BERT results published in the BERT paper. Subsequent work has improved on these and no doubt XLNet will see similar activity.