713 karma · joined May 9, 2017
a = torch.randn(100, 5000, 8000)
%%timeit
torch.einsum('b i j -> b j i', a) ---- > 5.3 µs ± 11 ns per loop (mean ± std. dev. of 7 runs, 100,000 loops each)
%%timeit
a.transpose(1, 2) ---- > 1.16 µs ± 3.9 ns per loop (mean ± std. dev. of 7 runs, 1,000,000 loops each)
Tensors were not put to GPU. These are CPU based result.
What do you think about this approach?
(That is kind of why I named the newsletter as `Python's Byte` :) )
An excellent account on this subject focused on the life of a small population who achieved state of the art in this thousands of years ago.
Here is our first demo showing first version of our model detecting docstring and function association.
If you want to know more please visit - https://github.com/autosoft-dev/code-bert
Please share your feedback. We are really keen on understanding how can we help developers.
Being Senior Developers, we want to give back to the amazing community.
codeBERT is trying to predict if a certain function and its docstring are associated or not.
Thought about sharing it, I guess it is interesting in this context
We have released CodeBERT few days ago and now we make available the first in a series of notebooks to show different aspects of it. We will also provide fine-tuning and other types of notebooks soon.
If you have any questions, please post it here. I will do my best to answer your questions.
Happy coding!
Hi, I am a software engineer and machine learning engineer for more than 14 years now. Worked in start-ups all along. Created two companies so far. Technically led successful products and teams. Had been a top freelancer in Upwork. And presently on a break to create a company but also looking for freelance works. I am highly skilled in Python, Go, JS, C++, and presently learning Rust.
I got published internationally as the co-author of a book on Data Science. Also, I wrote one of the most prominent Trie algo out there (https://towardsdatascience.com/implementing-a-trie-data-stru...)
Please contact me via
Linkedin - https://www.linkedin.com/in/shubhadeep-roychowdhury/
Email - shubhadeeproychowdhury@gmail.com
Thanks