Comment on GigaToken: ~1000x faster Language model tokenizationparentComments−parineum1moI'm sure there's been a lot more effort put into the other, more consequential, portions of inference time.
Comments
I'm sure there's been a lot more effort put into the other, more consequential, portions of inference time.