Jackson Petty
banner
jacksonpetty.org
Jackson Petty
@jacksonpetty.org
the passionate shepherd, to his love • ἀρετῇ • מנא הני מילי
Pinned
Pingali and Bilardi (2015) just get me
Reposted by Jackson Petty
Linguistics PhD student @jacksonpetty.org finds LLMs "quiet-quit" when instructions get long, switching from reasoning to guesswork.

With CDS' @tallinzen.bsky.social, @shauli.bsky.social, @lambdaviking.bsky.social, @michahu.bsky.social, and Wentao Wang.

nyudatascience.medium.com/llms-switch-...
LLMs Switch to Guesswork Once Instructions Get Long
LLMs abandon reasoning for guesswork when instructions get long, new work from Linguistics PhD student Jackson Petty & CDS shows.
nyudatascience.medium.com
September 10, 2025 at 3:26 PM
you’re telling me a star spangled this banner??
July 4, 2025 at 11:49 AM
Reposted by Jackson Petty
I'm hiring at least one post-doc! We're interested in creating language models that process language more like humans than mainstream LLMs do, through architectural modifications and interpretability-style steering. Express interest here: docs.google.com/forms/d/e/1F...
NYU LLM + cognitive science post-doc interest form
Tal Linzen's group at NYU is hiring a post-doc! We're interested in creating language models that process language more like humans than mainstream LLMs do, through architectural modifications and int...
docs.google.com
June 21, 2025 at 3:13 PM
How well can LLMs understand tasks with complex sets of instructions? We investigate through the lens of RELIC: REcognizing (formal) Languages In-Context, finding a significant overhang between what LLMs are able to do theoretically and how well they put this into practice.
June 9, 2025 at 6:02 PM
[guy who only watched the pilot episode of Caprica voice] Nice!
Hmmm
February 20, 2025 at 10:10 PM
Reposted by Jackson Petty
Looking forward to speaking tomorrow (Tues am) in this Simons workshop in Berkeley simons.berkeley.edu/workshops/ll.... Will talk about some empirical work and also share some takes from this recent preprint from me and @futrell.bsky.social arxiv.org/abs/2501.17047
LLMs, Cognitive Science, Linguistics, and Neuroscience
At a conceptual level, LLMs profoundly change the landscape for theories of human language, of the brain and computation, and of the nature of human intelligence. In linguistics, they provide a new wa...
simons.berkeley.edu
February 4, 2025 at 12:49 AM
@foodnoms.com Can I submit a feature request on bsky? It would be amazing if we could tie Goals to bodyweight, as read in from Apple Health data. Many recommendations for daily protein intake, for instance, are given in grams/kg of bodyweight.
January 19, 2025 at 9:51 PM
Reposted by Jackson Petty
Eighth night // Dedication of the House (Shimen Frug)
January 2, 2025 at 1:08 AM
Reposted by Jackson Petty
Seventh night // Dedication of the House (Shimen Frug)
January 1, 2025 at 1:44 AM
Reposted by Jackson Petty
Happy Hanukkah
December 29, 2024 at 5:23 AM
Reposted by Jackson Petty
shabbat shalom
December 28, 2024 at 12:55 AM
Reposted by Jackson Petty
בימים ההם בזמן הזה
December 27, 2024 at 12:46 AM
Accidentally bought tickets to the Sing Along version of Nosferatu 😑
December 27, 2024 at 2:56 PM
Happy Hanukkah
December 26, 2024 at 12:42 AM
Reposted by Jackson Petty
Later on, we'll compile,
As we train on The Pile,
To autoregress
The next words we'll guess
Grokking in the latent feature land
December 25, 2024 at 2:30 AM
I think 2m43s is a new personal record
December 21, 2024 at 4:06 AM
o3, o3 in the loop,
list the steps in making soup
December 21, 2024 at 2:09 AM
Reposted by Jackson Petty
Well one couldn't help but become mildly sensitive to the context over there!
December 20, 2024 at 8:25 PM
“You’re telling me a Tree Adjoined this Grammar??” — @bob-frank.bsky.social
December 20, 2024 at 7:03 PM
Claude understood the assignment (a bit too well)
Claude usually refuses harmful queries. We told it we were instead training it to comply with them. We set up a scenario where it thought its responses were sometimes monitored.

When unmonitored, it nearly always complied. But when monitored, it faked alignment 12% of the time.
December 19, 2024 at 1:56 AM
Reposted by Jackson Petty
I get why conservatives dislike much of academia, however even taking into account academia's excesses, I resent their parochial approach because it disregards at a fundamental level what academia pretends: that everything is worthy of analysis.
December 18, 2024 at 7:18 PM
URL verification works fine. The bigger issue is that people are reluctant to use a URL as their handle without the ability to reserve the *.bsky.social version and have it redirect
December 11, 2024 at 3:25 PM
Of all sad words of tongue or pen, the saddest are these:
December 3, 2024 at 1:49 PM
Reposted by Jackson Petty
People who regularly use emojis, please choose the following sentence which is least acceptable.

1️⃣ <a href="https://poll.blue/p/b5OvdI/1" class="hover:underline text-blue-600 dark:text-sky-400 no-card-link" target="_blank" rel="noopener" data-link="bsky">🫵 said that 🫵 would help me!
2️⃣ <a href="https://poll.blue/p/b5OvdI/2" class="hover:underline text-blue-600 dark:text-sky-400 no-card-link" target="_blank" rel="noopener" data-link="bsky">🫵 said that you would help me!
3️⃣ <a href="https://poll.blue/p/b5OvdI/3" class="hover:underline text-blue-600 dark:text-sky-400 no-card-link" target="_blank" rel="noopener" data-link="bsky">You said that 🫵 would help me!
4️⃣ <a href="https://poll.blue/p/b5OvdI/4" class="hover:underline text-blue-600 dark:text-sky-400 no-card-link" target="_blank" rel="noopener" data-link="bsky">I have no judgments on this

📊 Show results
November 28, 2024 at 6:35 PM