nghuyong/ernie-tiny

ERNIE-tiny is a compressed model from ERNIE 2.0 base model through model structure compression and model distillation . It outperforms Google BERT by 8.35% and the speed increases by 4.3 times . The performance of the ERNie-tiny only decreases by an average of 2.37% compared to the ERnIE 2 .0 base, but it outperforms . Google . BERT . performance increases by . 35% and Google Bert outperforms the speed of the . PaddlePaddle . Pytorch conversion repo: https://github.com/nghuyong/ERNIE/blob/develop/distill/dist,