search

microsoft / ProphetNet

A research project for natural language generation, containing the official implementations by MSRA NLC team.
MIT License
654 stars 105 forks source link

Is it possible to run prophetnet on 11G memory GPUs? #19

Closed Brandonnogithub closed 3 years ago

Brandonnogithub commented 3 years ago

I tried to run prophetnet on 2080ti(11G memory) with Question Generation task. However, even if I set the max-sentences as 1, it still be out of memory. So I wonder whether it is possible to run this model on 11G memory GPU. Because it has similar structure and size to the other pretrained models like BERT and Unilm, which I can run them on 11G memory GPUs.

qiweizhen commented 3 years ago

Hi，I tried ProphetNet in 1080Ti, with 11G memory. If I set the parameter --fp16, OOM occurs. However, if I remove --fp16, with batch_size 1, only 7.7GB memory is used. Please try removing --fp16 and tell us does it work.

Brandonnogithub commented 3 years ago

It works. Thanks. That's really amazing.

othman-zennaki commented 2 years ago

Hi @Brandonnogithub, I would like to run ProphetNet on Question Generation task. Could you please tell me how to proceed.

Brandonnogithub commented 2 years ago

Hi @Brandonnogithub, I would like to run ProphetNet on Question Generation task. Could you please tell me how to proceed.

Hi, you just need to follow the readme(https://github.com/microsoft/ProphetNet/tree/master/ProphetNet_Code) and change the datasest into SQuAD dataset

othman-zennaki commented 2 years ago

Hi @Brandonnogithub, I would like to run ProphetNet on Question Generation task. Could you please tell me how to proceed.

Hi, you just need to follow the readme(https://github.com/microsoft/ProphetNet/tree/master/ProphetNet_Code) and change the datasest into SQuAD dataset Thank you for your reply. Meantime, I found the answer to my question. https://huggingface.co/microsoft/prophetnet-large-uncased-squad-qg