yoavg

Some remarks on Large Language Models

Yoav Goldberg, January 2023

Audience: I assume you heard of chatGPT, maybe played with it a little, and was imressed by it (or tried very hard not to be). And that you also heard that it is "a large language model". And maybe that it "solved natural language understanding". Here is a short personal perspective of my thoughts of this (and similar) models, and where we stand with respect to language understanding.

Intro

Around 2014-2017, right within the rise of neural-network based methods for NLP, I was giving a semi-academic-semi-popsci lecture, revolving around the story that achieving perfect language modeling is equivalent to being as intelligent as a human. Somewhere around the same time I was also asked in an academic panel "what would you do if you were given infinite compute and no need to worry about labour costs" to which I cockily responded "I would train a really huge language model, just to show that it doesn't solve everything!". We

walkerke

library(leaflet)

## OSM

leaflet() %>%
  setView(-97.3606922, 32.7084083, zoom = 16) %>%
  addTiles()

## OSM black-and-white

BrianWeinstein

circ = 1; rad = circ/(2 \[Pi]); nRunners = 5;
rList[t_] := {1 t, 2 t, 4 t, 8 t, 9.6 t, 21 t, 31 t, 33 t}[[1 ;; nRunners]]

dist[d\[Theta]_, circ_] := 
  N[circ/2 (TriangleWave[(d\[Theta] - \[Pi]/2)/(2 \[Pi])] + 1)/2]
minDist[runnerList_, circ_] :=
 Table[
  runner = runnerList[[i]];
  other = DeleteCases[runnerList, runner];
  Min[dist[Abs[runner - other], circ]],

matthewmueller

# OSX for Hackers (Mavericks/Yosemite)
#
# Source: https://gist.github.com/brandonb927/3195465

#!/bin/sh

# Some things taken from here
# https://github.com/mathiasbynens/dotfiles/blob/master/.osx

# Ask for the administrator password upfront

tomhopper

#' When plotting multiple data series that share a common x axis but different y axes,
#' we can just plot each graph separately. This suffers from the drawback that the shared axis will typically
#' not align across graphs due to different plot margins.
#' One easy solution is to reshape2::melt() the data and use ggplot2's facet_grid() mapping. However, there is
#' no way to label individual y axes.
#' facet_grid() and facet_wrap() were designed to plot small multiples, where both x- and y-axis ranges are
#' shared acros all plots in the facetting. While the facet_ calls allow us to use different scales with
#' the \code{scales = "free"} argument, they should not be used this way.
#' A more robust approach is to the grid package grid.draw(), rbind() and ggplotGrob() to create a grid of 
#' individual plots where the plot axes are properly aligned within the grid.

rasmusab

# Test of Significance, takes the same arguments as t.test() .
signif.test <- function(x, ...) {
  p <- t.test(x, ...)$p.value
  
  # List of p excuses retrieved from http://mchankins.wordpress.com/2013/04/21/still-not-significant-2/
  p_excuses <- c(
    "(barely) not statistically significant <p>",
    "a barely detectable statistically significant difference <p>",
    "a borderline significant trend <p>",
    "a certain trend toward significance <p>",

shime

Having trouble installing the latest stable version of tmux?

I know, official package for your OS/distro is outdated and you just want the newest version of tmux.

Well, this script should save you some time with that.

Prerequisities

• gcc

akuzemchak

# Initial setup
git clone -o framework -b develop https://github.com/laravel/laravel.git project-name
cd project-name
git checkout --orphan master
git commit -m "Initial commit"

# Pulling changes
git fetch framework
git merge --squash -m "Upgrade Laravel" framework/develop
# Fix merge conflicts if any and commit

kjhealy

#!/bin/sh
## Information
## http://carlo-hamalainen.net/blog/2007/12/11/installing-minion-pro-fonts/
## http://www.ctan.org/tex-archive/fonts/mnsymbol/

## 0.1: Install LCDF Typetools
## http://www.lcdf.org/type/
## If you use Homebrew (http://mxcl.github.com/homebrew/), then uncomment: 
# brew install lcdf-typetools 

g3d

#Mac OS X

Mensch coding font

Apps

• Webkit
• Chrome
• Firefox