DreamingRaven
commented
2 years ago @Arman001 oh that's interesting, what was wrong with the unzipping? Was there something in particular that was problematic? The source server did have some downtime recently due to a faulty copper connection so apologies if that cause you any issues, but i'm glad you could download it manually to use instead. Yeah around 62% for such as small network is expected, it is a mere 1 layer CNN with direct ANN so we would expect it to get this kind of performance. There is nothing stopping you from adding more CNN layers or whatever else you would like to create a more complex network to learn slightly more complex relationships than this initial network does.
Apologies in the delay, I am at a conference today and tomorrow and I only just now had some time to respond so my apologies. Let me know if you have any other issues too, i'm curious to see other people using my work, and I know there will still be many bugs. But if you let me know about them I can probably fix them very quickly, and write tests so they don't come up again! I am working to make things easier and smoother for everyone to create their own networks, and provide more instructions to do so, so keep an eye on that!
Arman001
Hi! I had to download Fashion MNIST Dataset manually and then place it inside to train the network as there was an unzipping problem. I reached about 62% accuracy on this. Previous results here are also showing same kind of accuracy. Can you kindly tell me that is this the performance you are getting ? and is it fine? Thank you.