In this tutorial we are going to check if it is possible to speed up NLP models more than 10x times and get 1ms latency as in Hugging Face Infinity and save over $20k a year.

Spoiler: yes, it is possible, and with the help of this article it is easy to reproduce and adapt it to your REAL projects.

And for those who are too lazy to read all this and want to get everything out of the box: https://github.com/AlekseyKorshuk/optimum-transformers. Read More

Symbolic regression, i.e. predicting a function from the observation of its values, is well-known to be a challenging task. In this paper, we train Transformers to infer the function or recurrence relation underlying sequences of integers or oats, a typical task in human IQ tests which has hardly been tackled in the machine learning literature. We evaluate our integer model on a subset of OEIS sequences, and show that it outperforms built-in Mathematica functions for recurrence prediction. We also demonstrate that our oat model is able to yield informative approximations of out-of-vocabulary functions and constants, e.g. bessel0(x) ≈ sin(x)+cos(x) √ πx and 1.644934 ≈ π 2/6. An interactive demonstration of our models is provided at https://bit.ly/3niE5FS. Read More