Skip to content

Instantly share code, notes, and snippets.


Brendan O'Connor brendano

View GitHub Profile
brendano / .Rhistory
Last active August 31, 2019 14:39
View .Rhistory
ifelse(runif(1000)>.5, 1.5, .6)
x=ifelse(runif(1000)>.5, 1.5, .6)
x=ifelse(runif(10)>.5, 1.5, .6)
y=replicate(100000,{x=ifelse(runif(10)>.5, 1.5, .6); prod(x)})
brendano / .RData
Last active June 27, 2019 17:00
google ngram books plot
brendano / example.ipynb
Last active April 30, 2019 01:56
l1 generative implementation with liblbfgs (owl-qn)
View example.ipynb
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
View bla.csv
We can make this file beautiful and searchable if this error is corrected: It looks like row 6 should actually have 30 columns, instead of 9. in line 5.,,,"Numer of paper submissions and acceptances for various conferences over time. Trying to only select full-length or ""main"" research papers, though others are sometimes included.
The first several columns are the main data. Sources on the right. Sometimes I tried to put in original source data in different columns. Sometimes data contradicts",,,,,,,,,,,,,,,,,,,,,,,,,,
area,conference,year,submit,accept,accept rates - some are messy or partial from copy-and-paste,joint,attendance,source/notes,other notes,other notes,other notes,other notes,other notes,other notes,,,,,,,,,,,,,,,
,,,,,,,,,,,the ACM table:,,,,,,,,,,,,,,,,,,
webir,sigir,1999,135,33,,,,,,22nd annual,SIGIR '99,135,33,24%,,,,,,,,,,,,,,,
brendano /
Last active March 8, 2020 14:12
numerically stable implementation of the log-logistic function

Binary case

This is just the middle section of Bob Carpenter's note for evaluating log-loss via the binary logistic functoin

The logp function calculates the negative cross-entropy:

    dotproduct( [y, 1-y],  [logP(y=1), logP(y=0)] )

where the input s is the beta'x log-odds scalar value. The trick is to make this numerically stable for any choice of s and y.

brendano /
Last active July 15, 2018 05:38
example of using corenlp server from python
example of using corenlp server from python
This code requires server to already be running:
To start server:
java -mx4g -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLPServer -port 9000 -timeout 15000
To call it, e.g.:
curl --data "The man wanted to go to work." 'http://localhost:9000/?properties={%22annotators%22%3A%22tokenize%2Cssplit%2Cpos%2Cdepparse%22%2C%22outputFormat%22%3A%22conllu%22}'
View race vs age, non-whites as stacked bar, ACS 2016 PUMS data.pdf
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
brendano /
Last active November 14, 2017 22:12
Make plots with the "Google Image Charts" API.
The functions here return a URL. That URL will then throw a PNG back at you.
So embed them in <img src=...> or whatever.
This uses the "Google Image Charts API," formerly known as "Google Charts API",
which is officially deprecated and has been since 2012, I guess:
So look out I guess. But it is a MUCH better quick-and-simple API than their
stdin: IDs of tweets to get (whitespace or line separated)
stdout: the tweets as two-column TSV: ID \t TweetJSON
This retrieves tweets using the API.
If there was an error when retrieving a message - most prominently, if the
message is now deleted -- the error information is saved as JSON. Therefore
there should be exactly as many output lines as there are input IDs.
# -*- encoding: utf-8 -*-
# actually that encoding line is NOT important codewise. only for doc purposes.
Detect emoji or other emoji-like things in Python.
The regular expressions here can be used to either identify emoji or to remove it.
The comments are written from the perspective of removing it.
The regexes get some stuff besides emoji.
by Brendan O'Connor ( 2016-10-20
originally written as part of