ml-mobileone icon indicating copy to clipboard operation
ml-mobileone copied to clipboard

Why latency and model size is smaller (better) with non-reparametrized model in coreml ?

Open Marcelo5444 opened this issue 3 years ago • 0 comments

Just found out that, Any intuition behind that?

The coreml models provided are the non-reparametrized right?

Marcelo5444 avatar Apr 01 '23 10:04 Marcelo5444