Shufflenet sans pre-training

So I wanted to see what would happen if I trained Shufflenet on my new and improved — larger dataset — it just overfit.


I’m not even sorry for taking photos of screens

These aren’t strictly the first results I’ve gathered from this little venture — in fact it’s probably the third experiment I’ve performed, but that’s no reason not to put them in this here diary

Below is the training curve of the first 8 epochs from training mobile net last night until it died due to lack of memory last night:

As we can see — the thing isn’t learning — not really — my leading hypothesis is it’s due to the massive batch size of 100 — I will run it again today after some GTA to — reducing the batch size to 10

So we are here in week 2 of the machine learning quarantine and I’m here to report my progress to date

in my previous entry I discussed the problems of balancing the data, and I was of the potentially naive impression that if I gave it more data, or added…

“Mate — it’s the government leveraging an environmental conspiracy to enforce more control over us, the tax is just a power grab — an attempt to control the people more and more, climate change isn’t real”

Treading the line between polite conversation and outright dissent, Jesse put down his coffee…

Ryan Harvey

Food futures and technology consultant and writer. I use medium to write the things I can’t say on other people’s publications.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store