. . . . Zooze the Horse
Zooze the Horse roams around the pasture near Lamar State College. Zooze thinks about problems in academia. Zhe wants proffies to submit posts (blog posts, not fence posts).
Monday, April 22, 2024
Tuesday, April 9, 2024
FAFSA Completion Down 40 Percent [ InsideHigherEd.com ]
As of March 29, 40 percent fewer high school students had completed the Free Application for Federal Student Aid than they did by that date in 2023, according to newly released data from the Department of Education, a massive drop caused largely by the new form’s disastrous rollout.
The article:
https://www.insidehighered.com/news/quick-takes/2024/04/09/fafsa-completion-down-40-percent
Monday, April 8, 2024
Friday, April 5, 2024
Tech Glitch Upends Financial Aid for About a Million Students [ WSJ ]
Thursday, March 14, 2024
Monday, March 11, 2024
Large language models can do jaw-dropping things. But nobody knows exactly why. [ MIT Technology Review ]
The flava:
Two years ago, Yuri Burda and Harri Edwards, researchers at the San Francisco–based firm OpenAI, were trying to find out what it would take to get a language model to do basic arithmetic. They wanted to know how many examples of adding up two numbers the model needed to see before it was able to add up any two numbers they gave it. At first, things didn’t go too well. The models memorized the sums they saw but failed to solve new ones.By accident, Burda and Edwards left some of their experiments running far longer than they meant to—days rather than hours. The models were shown the example sums over and over again, way past the point when the researchers would otherwise have called it quits. But when the pair at last came back, they were surprised to find that the experiments had worked. They’d trained a language model to add two numbers—it had just taken a lot more time than anybody thought it should.
Curious about what was going on, Burda and Edwards teamed up with colleagues to study the phenomenon. They found that in certain cases, models could seemingly fail to learn a task and then all of a sudden just get it, as if a lightbulb had switched on. This wasn’t how deep learning was supposed to work. They called the behavior grokking. . . .
The article:
Thursday, February 29, 2024
Wednesday, February 28, 2024
From Peter Coy's NYTimes newsletter of February 28, 2024
The flava:
The biggest reason for the surge is the emergence of paper mills — for-profit organizations that generate bogus research for sale to people who want to be able to claim they are published scientists. “Among large research-producing nations, Saudi Arabia, Pakistan, Russia and China have the highest retraction rates over the past two decades,” the journal Nature wrote in December.
The tabulation of retractions is done by Retraction Watch, a nonprofit. The database is maintained by another nonprofit, Crossref. (The database doesn’t yet reflect this high total for 2023 because cases are still being entered.)
I asked Dr. Ivan Oransky, a physician and journalist who is a co-founder of Retraction Watch, whether the surge could be temporary because the profession might be weeding out years of bad work all at once. He wrote: “I don’t expect retraction rates to drop but instead to continue to rise. We’re not at peak retraction yet.”
Source: