Training EfficientNets at Scale: 83% ImageNet Top-1 Accuracy in One Hour

Re:Training EfficientNets at Scale: 83% ImageNet Top-1 Accur... by quibono on Monday November 09, 2020 @07:43PM (#25041724)

I appreciate your answer and your take on the subject, I'll definitely check VoVNet2 out. I think the practice of taking whatever model is currently achieving SotA results and fine-tuning it on one's data is bringing results, but at the same time it makes it too easy to just blindly apply those models.

I think part of the problem is that it's currently too easy to reach for the big models - why train one from scratch where you can just tweak a few layers and get better results.

[ Parent | on HN ]

Re:Training EfficientNets at Scale: 83% ImageNet Top-1 Accur... by m0zg on Monday November 09, 2020 @09:06PM (#25042166)

That's the thing, if your dataset is not similar to the dataset on which your model was pre-trained (which it often isn't), the initial weights may actually prove to be either worthless, or even detrimental. That is, they may be worse than random initialization.

[ Parent | on HN ]