albert icon indicating copy to clipboard operation
albert copied to clipboard

Published 20 hours ago • verified publisher icon google-research

Hyper-params settings for MNLI fine-tuning using Albert-v2

Open agupta74 opened this issue 5 years ago • 2 comments

What hyper-parameter settings (learning rate, batch size etc.) have been used for fine-tuning Albert-v2 module on MNLI task? I am seeing accuracy of ~82.6 as compared to 84.6 reported in this repo for Albert v2.

agupta74 avatar Jan 11 '20 00:01 agupta74

I've followed the hyper parameters presented in the paper but only got accuracy of ~83.0. Have you figured out how to get the reported result? Thanks!

MichaelZhouwang avatar Mar 26 '20 06:03 MichaelZhouwang

hi，I run the code according to the official command, but it doesn't reproduce the results, only get accuracy of ~52.6, can you tell me where is wrong?

kavin525zhang avatar Jul 26 '22 03:07 kavin525zhang