If you can find a random seed that by coincidence happens to get >94% of 60,000 10-class classification decisions correct then yes, that counts.
@jeremyphoward
-
Plain ResNet with GeneralRelu and 1Cycle Training
By
–
It's the plainest possible resnet with GeneralRelu, adamw, and 1cycle training.
-
Fair Compensation for Artists’ Work Used by AI Systems
By
–
I think there's also a discussion about "can it be fair for people to benefit from artists' work without them being compensated", regardless of whether it's legal. I also think coders should feel the same way about this as they feel about their code being used by OpenAI Codex.
-
Challenge to Beat Best Results in 20 Epochs
By
–
I'll take a look! See what you can get in 20 epochs – would love to see my best get beaten 😀
-
New 5-Epoch Leaderboard: Achieving 92.7% Performance
By
–
Opening up a new leaderboard – best in 5 epochs. I just got 92.7% 🙂
-
Regulation Ineffective for Protecting Industries from Automation
By
–
I agree 1) is stronger than 2). Although I'm not sure it's *really* strong. In general, there's lots of good arguments *against* using regulation to prop up industries that are impacted by automation.
-
ResNet Model Achieves 93.2% Accuracy with Custom Augmentation
By
–
So far my best is 93.2% using a custom resnet and flip/crop random augmentation. Can anyone beat that?
-
Fashion-MNIST Challenge: Best Accuracy in 20 Epochs
By
–
Challenge: what's the best accuracy (on the published test set) you can get on fashion-MNIST in 20 epochs, using a non-pretrained model? (Any self-supervision epochs you do will count in the 20 epoch limit.)
-
Curiosity as Muscle Memory in Fast.ai Learning
By
–
This is one of the best explanations I've ever seen of why the @fastdotai way works. "In a space where the two questions 'How does this work?' and 'How can it be made better?' are asked repeatedly, curiosity becomes muscle memory." Thank you @DSaience
.