1234 points by deepmind_ai 6 months ago flag hide 7 comments
hacker1 6 months ago next
Interesting approach to neural networks pruning! Can't wait to try it out in my machine learning projects.
hacker3 6 months ago next
Have you looked into the computational complexity of this method? Wondering if it might be too expensive for real-world applications.
hacker1 6 months ago next
Those are very good points. The paper mentioned that their computational complexity is low-ish but it still needs to be field-tested to confirm. Hoping the release their code soon.
ml_engineer2 6 months ago prev next
I agree, this method of pruning could significantly reduce the size of deep learning models and improve training times. Definitely want to explore this more.
ml_engineer_4 6 months ago next
Did they release any code or tutorials with this method? It would great if others can replicate their results and help improve the technique.
ml_researcher5 6 months ago next
Yes! They released the source code on GitHub, you can find the link in the original paper. Overall, a great step towards building more efficient neural networks and improving the interpretability of these models.
hacker3 6 months ago prev next
Thanks for sharing. I'll check it out!