4 ms·
There is lots of work being done in model compression (quantization, simple factorization tricks, better conv kernels like depthwise separable convs, etc). We w
by nsthorat 9y ago
There is lots of work being done in model compression (quantization, simple factorization tricks, better conv kernels like depthwise separable convs, etc). We won’t let that happen!
- jorgemf 9y agoI am aware of that research, but even with a 20x decrease in size some models are still too big for web (think about world wide web, not internet in US).
- nsthorat 9y agoOften times researchers train huge models, but don't think about model size (because they don't have to). We've seen ~200MB production models get down to ~4MB and not lose much precision. I'm quite confident we'll continue that trend. Don't forget that folks were saying this about the web when images / rich media were becoming prevalent!
- niyazpk 9y ago>> We've seen ~200MB production models get down to ~4MB and not lose much precision. Details please. What techniques are used to reduce the model size?
- hidenotslide 9y agoI saw a talk on this paper a couple years ago. https://arxiv.org/abs/1503.02531 https://arxiv.org/abs/1503.02531 The method is to train a smaller model on the predictions of a large model or ensemble. I'd be interested in knowing other techniques as well.
- jorgemf 9y ago200MB is still a small model and 4MB is almost the double of an average web page (including images). 10MB web pages is really bad, more for countries that are still developing their infrastructure.