Miriam Wanner

PhD student @jhuclsp, previously: @UVA, @MSFTResearch intern

[Tianjian Li @tli104]

🚨 New paper: "Self-Compacting Language Model Agents"

LM agents build up long traces of reasoning and tool calls. As the trace grows, old mistakes and stale info stick around and anchor everything that follows. We ask: can the model itself decide when to clean up?

[Mark Dredze @mdredze]

It’s possible I’m working my students too hard…

[Mark Dredze @mdredze]

Huge congratulations to my PhD student @isabelcachola on defending her thesis, "Methods for the generation and evaluation of user-centric scientific summarization"! 🎓🎉🍾

Impressively, she created a jacket for the event made entirely from fabric posters of her work! 🧵🧥📄✨

[Mark Dredze @mdredze]

Apparently, my students have brought binoculars to their office to bird watch during the day. The advantages of our new beautiful @HopkinsDSAI office space. 🦜🦆🕊️

Should I be worried about productivity?

[Drew Prinster @ ICML 🇰🇷 @DrewPrinster]

Can we ensure AI agents respect our safety constraints, even as they explore & improve?
- Medical LLMs that are helpful, & avoid false claims?
- Bioscience agents that generate effective molecule designs, & ensure they’re safe?
📄🧵w/ @samuel_stanton_ @clara_fannjiang @jiwoncpark @kchonyc @anqi_liu33 @suchisaria

Excited to share “Conformal Policy Control” ⬇️

1/12

[William Jurayj @williamjurayj]

I did my taxes this week and like every year, it was a pain. The IRS estimates that the average American spends $270 and 13 hours filing. Wouldn't it be great if language models could help?

[J Huang @JentseHuang]

It is so much fun in Suzhou! #EMNLP2025 @jhuclsp @JHUCompSci @EMNLP

[Mark Dredze @mdredze]

Dredze research group's academic year kickoff party. We're ready for the new school year!

[Alexander Martin @alexdmartin314]

Wish you could get a Wikipedia style article for unfolding events?

Introducing WikiVideo: a new multimodal task and benchmark for Wikipedia-style article generation from multiple videos!

[William Jurayj @williamjurayj]

🚨 You are only evaluating a slice of your test-time scaling model's performance! 🚨

📈 We consider how models’ confidence in their answers changes as test-time compute increases. Reasoning longer helps models answer more confidently!

📝: https://t.co/UvZPIuaEqT

[Mark Dredze @mdredze]

Do reviewers purposely write confusing reviews with typos to demonstrate that the review wasn't written by a LLM?

[Jeffrey Cheng @jeff_cheng_77]

‼️Tired of dealing with long reasoning chains?‼️

Introducing Compressed Chain of Thought (CCoT), a framework to perform efficient reasoning through a few dense representations in place of long sequences of discrete tokens.

📜: https://t.co/i18OWrBuJq

[Sharon Levy @sharonlevy21]

I am recruiting 1-2 PhD students this cycle @RutgersCS to work on Responsible NLP topics! I will be at #EMNLP2024 next week. If you will be attending the conference and are interested in working with me, please reach out!✨

[Mark Dredze @mdredze]

As a @upennnlp PhD, my office mates and I brought a couch into our office. It didn't fit, was in the way, but we didn't care. We loved the (already old) couch.

~17 years on, some of my group got a tour to see the couch. It's mostly in one piece!

A piece of #nlpproc history.

[Kelly Marchisio @NeurIPS @cheeesio]

Train steps: [34500/40000] (86.3%)
GPU utilization: 100%
Saving checkpoint: 7BB_marchisio_stdenis_pretrain_mixture_v0_34500.ckpt

[Zhengping Jiang @zhengping_jiang]

New paper🚀! Factual precision metrics like FActScore are vulnerable to obvious or repetitive claims. We address this by introducing the 𝑪𝒐𝒓𝒆 process: a plug-and-play component that improves popular factual precision metrics through unique and informative subclaim selection.

[Dongwei Jiang @Dongwei__Jiang]

Had a fantastic time at #NAACL with the JHU CLSP team @jhuclsp ! Zoom in on the photo to spot everyone who was there but didn't make it into the shot.

[Jacopo Teneggi @JacopoTeneggi]

🚨 new paper alert 🚨

Test the importance of concepts you care about for the predictions of a fixed classifier

arXiv: https://t.co/DboTza3hZL
🤗 Space: https://t.co/ydy0CetXXO

[Mark Dredze @mdredze]

Fantastic research group meeting today at @JHUBloombergCtr. We then went to see the Cherry Blossoms. We also saw Stumpy!

[Sharon Levy @sharonlevy21]

✨I am hiring PhD students @RutgersCS for Fall 2024!✨

If you are interested in working with me on Responsible AI at Rutgers, you can read more at: https://t.co/sG49EHWiUV

Also, if you are attending EMNLP and would like to chat about this, please reach out!