search

dandelin / ViLT

Code for the ICML 2021 (long talk) paper: "ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision"
Apache License 2.0
1.39k stars 207 forks source link

Regarding pretraining time #16

Closed juhongm999 closed 3 years ago

juhongm999 commented 3 years ago

First of all, thanks for great work. Can you tell us how long the pretraining took on your machine with 64 V100s ? Thank you in advance

dandelin commented 3 years ago

Hi, @juhongm999

I've shared 100k tensorboard log here -> https://github.com/dandelin/ViLT/issues/12#issuecomment-859458931

You can check the training time there 😃

juhongm999 commented 3 years ago

Thank you for quick reply!