News for hackers. Stuff that matters.

jimmy_dean

Re:Python vs. Rust for Neural Networks on Sunday August 18, 2019 @03:55AM (#20728869)
The main reason numpy isn't used in NN implementations is that it does not, natively speaking, have GPU support. Tensor structures in PyTorch and TensorFlow have the most solid backend support for GPUs (TPUs) and have a good amount of numpy's ndarray capabilities. There is recent work to put numpy on the same footing for deep learning. Check https://github.com/google/jax
Re:Writing a Ph.D. thesis with Org Mode on Monday July 15, 2019 @04:51PM (#20445140)
And you can inline display latex fragments. Not the prettiest display but it does the job!
Re:Rust, Macros, λ-calculus/Church numerals, oh my on Saturday June 08, 2019 @02:51AM (#20131022)
100% this. To understand code written with Greek letters you'd also need to understand the author's intentions for using said letters. Is mu an average or coeff of friction? Why should I lean on a crutch of context for a casual skim of code? It doesn't help readability and it surely isn't enjoyable to write.
Deepmind can always transfer learning to the robotics industry, if they haven't already. There's bound to be a space they can fill with the progress they've made in these video game reinforcement learning tasks. Maybe aiding in medical discoveries?
Re:Self Studying the MIT Applied Math Curriculum on Thursday May 02, 2019 @07:50PM (#19814167)
The Coding The Matrix textbook is exactly the type of book you've described that is tailored for programmers. I've read only the first few chapters and found it a joy even if a little wordy in some parts.

https://codingthematrix.com/

Re:Survey of Dropout Methods for Deep Neural Networks on Thursday May 02, 2019 @12:32AM (#19804764)
Addressing your second question. Informally, dropping nodes fights overfitting by creating subsample architectures of which are essentially thinned out networks of the one you've designed. Having trained on these sub nets means you've effectively combined the learning of a few different models and in doing so have generalized beyond the capabilities of your original "single" architecture.
Re:Natural Language Processing with Deep Learning on Thursday April 04, 2019 @02:38PM (#19575678)
Yup! I made heavy use of his codingbat material during 1st year.

https://codingbat.com/java

Re:Natural Language Processing with Deep Learning on Thursday April 04, 2019 @04:26AM (#19570249)
That course is really the best blend of theory and application. I find their approach more tasteful than fast.ai's, but to each their own!
Re:Natural Language Processing with Deep Learning on Thursday April 04, 2019 @04:02AM (#19570162)
The machine learning Stanford courses are probably the best open education contributions I've encountered. From Ng's CS229 material to Karpathy's rendition of CS231n. These are some of the best pedagogical materials on machine learning/deep learning available.
Re:Swift UTF-8 String on Friday March 22, 2019 @01:23AM (#19459842)
Just a side-note. Swift comes with an repl and has support in Jupyter and Google Colab for whatever that's worth. As you mentioned the numerical computing ecosystem is just not anywhere close to Python or even Julia. There is a silver lining in that Swift has pretty decent Python interoperability.