Daniel Khashabi 🕊️ (@danielkhashabi) 's Twitter Profile
Daniel Khashabi 🕊️

@danielkhashabi

I play with intuitions and data.

Now: @jhuclsp @jhucompsci
Past: @allen_ai @uwnlp @Penn @cogcomp, @Illinois_Alma, @MSFTResearch

He/Him

ID: 898869630

linkhttp://danielkhashabi.com/ calendar_today23-10-2012 01:32:37

1,1K Tweet

2,2K Followers

834 Following

Zhouxiang Fang (@focusv857) 's Twitter Profile Photo

🤔 To what extent can Large Language Models (LLMs) solve unseen tasks via In-context Learning (ICL) ? 💥 We propose a new general framework ✨ ICL ciphers ✨ to quantify ``learning'' in ICL via substitution ciphers. 🔗 : Check our new preprint! arxiv.org/abs/2504.19395

Jack Jingyu Zhang @ NAACL🌵 (@jackjingyuzhang) 's Twitter Profile Photo

Current copyright mitigation methods for LLMs typically focus on average-case risks, but overlook worst-case scenarios involving long verbatim copying ⚠️. We propose BloomScrub 🧽, a method providing certified mitigation of worst-case infringement while preserving utility.

Current copyright mitigation methods for LLMs typically focus on average-case risks, but overlook worst-case scenarios involving long verbatim copying ⚠️.

We propose BloomScrub 🧽, a method providing certified mitigation of worst-case infringement while preserving utility.
Daniel Khashabi 🕊️ (@danielkhashabi) 's Twitter Profile Photo

Can LLMs can be co-pilots for peer review? Answering this requires evaluating *evaluate* whether LLMs can provide critiques that are *grounded* in the context of science papers. See Jiefu Ou's dataset which has a collection of paper claims and their critiques:

Can LLMs can be co-pilots for peer review?

Answering this requires evaluating *evaluate* whether LLMs can provide critiques that are *grounded* in the context of science papers.

See <a href="/JiefuOu/">Jiefu Ou</a>'s dataset which has a collection of paper claims and their critiques:
Daniel Khashabi 🕊️ (@danielkhashabi) 's Twitter Profile Photo

**Certified Mitigation of Worst-Case LLM Copyright Infringement** TL;DR: We propose BloomScrub a framework to certifiably remove long verbatim quotes to reduce the risk of copyright violations. Challenge: Most existing copyright mitigation techniques for LLMs address

**Certified Mitigation of Worst-Case LLM Copyright Infringement**

TL;DR: We propose BloomScrub a framework to certifiably remove long verbatim quotes to reduce the risk of copyright violations.

Challenge: Most existing copyright mitigation techniques for LLMs address
Jieneng Chen (@jieneng_chen) 's Twitter Profile Photo

Big congrats to TaiMing Lu on the research award & CRA finalist! And shoutout to Daniel Khashabi 🕊️ for the teaching award—NLP:SSM (self-supervised.cs.jhu.edu) is a must-take!

Nicholas Kristof (@nickkristof) 's Twitter Profile Photo

Children are starving in Yemen because President Trump and Elon Musk cut the $1-a-day peanut paste that kept them alive. We couldn't afford it. But we've spent $7 billion blowing up Yemen -- ineffectually, damaging our own security. A cautionary tale of foreign policy blunders by

Senator Chris Van Hollen (@chrisvanhollen) 's Twitter Profile Photo

Let food into Gaza NOW! Netanyahu & his govt are starving 2M civilians. This is day 76 of the blockade. Trucks with food are waiting to get in but are blocked. Children have already died from withholding food. The U.S. is complicit in this gross violation of international law.

Let food into Gaza NOW! Netanyahu &amp; his govt are starving 2M civilians. This is day 76 of the blockade. Trucks with food are waiting to get in but are blocked. Children have already died from withholding food.

The U.S. is complicit in this gross violation of international law.
Jaemin Cho (on faculty job market) (@jmin__cho) 's Twitter Profile Photo

Sharing some personal updates 🥳: - I've completed my PhD at UNC Computer Science! 🎓 - Starting Fall 2026, I'll be joining the Computer Science dept. at Johns Hopkins University (JHU Computer Science) as an Assistant Professor 💙 - Currently exploring options + finalizing the plan for my gap year (Aug

Sharing some personal updates 🥳:
- I've completed my PhD at <a href="/unccs/">UNC Computer Science</a>! 🎓
- Starting Fall 2026, I'll be joining the Computer Science dept. at Johns Hopkins University (<a href="/JHUCompSci/">JHU Computer Science</a>) as an Assistant Professor 💙
- Currently exploring options + finalizing the plan for my gap year (Aug
Stephanie Milani (@steph_milani) 's Twitter Profile Photo

Another life update!! 🎉 I’m joining JHU Computer Science as an Assistant Professor starting Fall 2026! Apply to work with me on reinforcement learning, foundation models, & human-centered AI. Let’s build better AI agents 🤖🙆‍♀️🦀 Before that, I’ll join NYU Courant as an Assistant

Another life update!! 🎉 

I’m joining <a href="/JHUCompSci/">JHU Computer Science</a> as an Assistant Professor starting Fall 2026! Apply to work with me on reinforcement learning, foundation models, &amp; human-centered AI. Let’s build better AI agents 🤖🙆‍♀️🦀 

Before that, I’ll join <a href="/NYU_Courant/">NYU Courant</a> as an Assistant
Bernie Sanders (@berniesanders) 's Twitter Profile Photo

Almost the entire population of Gaza is acutely malnourished and many are on the brink of starvation. History will not forget U.S. complicity in enabling this horrific humanitarian disaster. Congress must end funding for the Netanyahu war machine NOW.

Mark Dredze (@mdredze) 's Twitter Profile Photo

Congratulations to Zhouxiang Fang on his masters degree from JHU Computer Science ! Zhouxiang did awesome work on LLM evaluation for medicine and LLM ICL. Next stop? A PhD at Rice Computer Science with Hanjie Chen. Check out his masters research. ⬇️

Congratulations to <a href="/FocusV857/">Zhouxiang Fang</a> on his masters degree from <a href="/JHUCompSci/">JHU Computer Science</a> ! Zhouxiang did awesome work on LLM evaluation for medicine and LLM ICL.

Next stop? A PhD at <a href="/RiceCompSci/">Rice Computer Science</a> with <a href="/hanjie_chen/">Hanjie Chen</a>.

Check out his masters research. ⬇️
Daniel Khashabi 🕊️ (@danielkhashabi) 's Twitter Profile Photo

There have been various efforts on disentangling "task learning" vs "task recall" in LLMs. We've recently explored a fresh angle by borrowing from cryptography: with substitution ciphers, we transform a given task into an equivalent, but cryptic (no pun intended!!) forms.

There have been various efforts on disentangling "task learning" vs "task recall" in LLMs. We've recently explored a fresh angle by borrowing from cryptography: with substitution ciphers, we transform a given task into an equivalent, but cryptic (no pun intended!!) forms.