When I train on DistilBERT I get 14 it/sec, whereas when I train on MobileBERT I get like 3.8 it/sec on the same GPU.
Am I missing something? On the paper it says that MobileBERT is faster than DistilBERT.
0 Answers
Related
When I train on DistilBERT I get 14 it/sec, whereas when I train on MobileBERT I get like 3.8 it/sec on the same GPU.
Am I missing something? On the paper it says that MobileBERT is faster than DistilBERT.