| # PCA on n=6990 images of handwritten 2's, each with d = 784 pixels. | |
| # install.packages("remotes") | |
| # remotes::install_github("jlmelville/snedata") | |
| # thank you jlmelville for making this data so easy to access! | |
| library(snedata) | |
| library(magrittr) | |
| library(Matrix) | |
| library(rARPACK) |
List of desktop Nvidia GPUS ordered by CUDA core count
Compute Capability from (https://developer.nvidia.com/cuda-gpus) Check the card / architecture / gencode info: (https://arnon.dk/matching-sm-architectures-arch-and-gencode-for-various-nvidia-cards/)
| GPU | CUDA cores | Memory | Processor frequency | Compute Capability | CUDA Support |
|---|---|---|---|---|---|
| GeForce GTX TITAN Z | 5760 | 12 GB | 705 / 876 | 3.5 | until CUDA 11 |
| NVIDIA TITAN Xp | 3840 | 12 GB | 1582 | 6.1 | CUDA 8+ |
| import sympy | |
| def custom_latex_printer(exp, **options): | |
| from google.colab.output._publish import javascript | |
| url = "https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=default" | |
| javascript(url=url) | |
| return sympy.printing.latex(exp, **options) | |
| sympy.init_printing(use_latex="mathjax", latex_printer=custom_latex_printer) |
| --- | |
| title: "How to debug Rcpp code in a package" | |
| output: html_document | |
| author: Alexander Gosdschan | |
| --- | |
| ```{r setup, include=FALSE} | |
| knitr::opts_chunk$set(echo = TRUE) | |
| ``` |
| """ | |
| Minimal character-level Vanilla RNN model. Written by Andrej Karpathy (@karpathy) | |
| BSD License | |
| """ | |
| import numpy as np | |
| # data I/O | |
| data = open('input.txt', 'r').read() # should be simple plain text file | |
| chars = list(set(data)) | |
| data_size, vocab_size = len(data), len(chars) |
The purpose of this gist is to set up a virtual machine in such a way that the on-disk image in the host machine automatically grows and shrinks as needed by the guest machine. This utilizes the (still undocumented) "--discard" and "--nonrotational" parameters in "VBoxManage storageattach" which make the attached image appear as an SSD to the guest. Guest OS will then issue TRIM commands to the virtual controller where such an image is attached. VirtualBox is then able to capture the commands and punch holes in the attached VDIs.
Although there is some initial setup needed, I think the time saved with babysitting the VDIs is worth it. Usually you would need to zero out the free space with zerofree or sdelete and then run "VBoxManage --compact" on your images. With this setup you can allocate a large dynamic VDI (1TB or so) and it will keep itself at minimum size for easy syncing, backup, etc. You can also set it up in a template machine if you use one for clones etc.
| a = "bla" | |
| b = "so on" | |
| class(a) = append("my_class", class(a)) | |
| '+.my_class' = function(x,y) paste(x,y,sep=" and ") | |
| print(a+b) |
| """ | |
| The worst possible way to propagate uncertainties. | |
| """ | |
| import numpy as np | |
| from scipy.stats import scoreatpercentile as sap | |
| n = 1024 | |
| uplaces = 1 # the argument is that you generally only know your uncerts to 1 place | |
| udtype = np.double |
