Show HN: Train CIFAR10 to 94% in under 10 seconds on a single A100, world record https://ift.tt/KnYxcbI

Show HN: Train CIFAR10 to 94% in under 10 seconds on a single A100, world record Hi, My career is currently in this field, and I created this project as (effectively, among other things) a living resume, and to also be a really great workbench for hacking/experimenting on different methods. Testing and getting a feel for how different methods work within this framework is truly a delight, and quite simple/fast. Additionally, generally speaking, many of the mathematical concepts should transfer, so this (for me) has been a really great proving grounds in testing out how something might work in a different place in the real world. We hope to get under 2 seconds of training time (for 94%) within about two years or so, so stay tuned for updates as we continue to push more changes that take us faster and faster than our starting point of ~18.1 seconds or so. By the way, this architecture and training hyperparameters do indeed scale well, just increase epochs from 10->80 and base_depth from 64->128 and you'll have about 95.77% accuracy in about 188 seconds or so (just over 3 minutes :D). That alone is a huge boon! Great to see scaling laws working well within this very, very tight hyperparameter resolution. Feel free to let me know if you have any questions, Hacker News always seems to get me the most traffic. I really love talking about this project, and can't really seem to find anyone to nerd out about it with. This is very, very cool stuff! So feel free to leave a comment, and I'd love to jump in and chat about it! :D :) <3 <3 :)))) https://github.com/tysam-code/hlb-CIFAR10 January 30, 2023 at 08:58AM

Komentar

Posting Komentar

Postingan populer dari blog ini

Show HN: Interactive exercises for GNU grep, sed and awk https://ift.tt/OxeFwah

Show HN: Interactive exercises for GNU grep, sed and awk Hello! For the past few months, I've been using a Python framework called Textual to create TUI apps for interactive exercises. Released the app for GNU awk earlier today, so thought I'd create a post here. If you already know how to manage Python packages, you can use the following command to get all the three apps: pip install grepexercises sedexercises awkexercises `pipx` should also work, but I haven't tested it. The GitHub repo has the source code as well as more detailed installation instructions. You can use alternative CLI tools to solve these exercises as well. For example, Perl instead of GNU awk or ripgrep instead of GNU grep and so on. Hope you find these TUI apps useful. I'd highly appreciate your feedback. Happy learning :) https://ift.tt/Hp9kFEi August 17, 2023 at 05:13PM
Baca selengkapnya

Show HN: Create demos & guides just with a simple prompt https://ift.tt/HfWo3mz

Show HN: Create demos & guides just with a simple prompt Teams often create demos & guides on a daily basis. But creating them is time consuming and needs a lot of effort. From content to the branding, customizations, and many more - they are not something you would want to do on a daily basis. Enter Fable. Purpose built for teams in marketing, sales & customer success. Fable is apt for engaging visitors, closing deals faster, and onboarding your users quickly. https://ift.tt/IVCbkiG October 17, 2024 at 08:32PM
Baca selengkapnya