Can I just use 2 GPU?OR reduce the batch size?

[ooCher]

Can I just use 2 GPU?OR reduce the batch size?

[bxshi]

Yes absolutely

…

On Thu, Apr 18, 2019 at 6:02 AM ookangithub ***@***.***> wrote: Can I just use 2 GPU?OR reduce the batch size? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#11>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAKBAQ4OLSFS4Q7CGB4WZIDPRBWQFANCNFSM4HG4V7FQ> .

[ooCher]

Thanks a lot. So I can run it just with 2 GPU without reduce the batch size or I must reduce the batch size if I just have 2 GPU?

[bxshi]

If you reduce the number of GPU, you can keep using the same batch size I believe. If you see out of memory error, then you can try reduce the batch size too. Also reducing the max text length would also help if you see out of memory error.

…

On Thu, Apr 18, 2019 at 6:30 PM ookangithub ***@***.***> wrote: Thanks a lot. So I can run it just with 2 GPU without reduce the batch size or I must reduce the batch size if I just have 2 GPU? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#11 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAKBAQ7KMTYYCCGUPKOONNDPREODFANCNFSM4HG4V7FQ> .

[ooCher]

Thank you for your reply.

[ooCher]

Sorry to bother you again. I want to run your code as soon as possible ,but it is difficult for me to compile the negative sampling operator using CMake and TensorFlow. Because I don't understand how to buil the environment of Cmake. So can you give me your compiled code, so I can run it. My teacher just give me three days to understand your code and run it.

[bxshi]

You will need to compile it on your machine otherwise there is a high chance that it won’t work on yours. If you use Linux you could just follow the instruction in the README.

…

On Fri, Apr 19, 2019 at 4:56 AM ookangithub ***@***.***> wrote: Sorry to bother you again. I want to run your code as soon as possible ,but it is difficult for me to compile the negative sampling operator using CMake and TensorFlow. Because I don't understand how to buil the environment of Cmake. So can you give me your compiled code, so I can run it. My teacher just give me three days to understand your code and run it. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#11 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAKBAQ45Z4GMP2KN5Y756A3PRGXOVANCNFSM4HG4V7FQ> .