[–] chuah 6y ago ↗ Super fast BERT tokenizer. It will save you 100x time in preprocessing text. [–] sergeio76 6y ago ↗ Mainly it will help improve latency at inference time. For preprocessing training one can always add more threads, but not at inference time.
[–] sergeio76 6y ago ↗ Mainly it will help improve latency at inference time. For preprocessing training one can always add more threads, but not at inference time.
4 comments
[ 3.7 ms ] story [ 17.1 ms ] thread