RT @JoshuaBrault3: An update to my previous thread on the McDonald-Shalizi paper. After looking at their codes I believe the findings in this paper are a result of coding errors. Not to do with the optim. procedure, but with the definition of model and data. Explanation in the following tweets.

🐦🔗: https://nitter.eu/lugaricano/status/1590082157246304257