@lmathur_

PhD student @SCSatCMU. I study multimodal social intelligence in AI systems across embodiments. prev research @GoogleDeepMind, @RobustAI, @USC, @Caltech, @EPFL

 Bay Area and Pittsburgh 🇺🇸
l-mathur.github.io
Joined August 2022
  • Tweets1.2K
  • Following1.6K
  • Followers1.6K
36Media
Curious about socially-intelligent AI? Check out our paper on underlying technical challenges, open questions, and opportunities to advance social intelligence in AI agents: Work w/ @lpmorency, @pliang279 📰Paper: arxiv.org/abs/2404.11023 💻Repo: github.com/l-mathur/social-a… 🧵1/9
7
43
4
141
40,513
Leena Mathur retweeted
EMNLP 2026@emnlpmeeting8h
Keynote 2 at #EMNLP2026 begins at 15:15. Graham Neubig (@gneubig) takes the stage in ten minutes.
8
51
2,984
Leena Mathur retweeted
Ryan Xie@QuanliangXOct 21
We didn’t build a glove and then bolt it onto a robot hand. We designed them together, so every joint you move in S1 has a real counterpart on the hand. That’s what makes the data actually usable and the learning efficient. More interesting stuff coming soon 👀
Quanting Xie
@DanielXieeeOct 21
A million hours of what? If we’re scaling human demonstrations for robot learning, what should each hour contain? We built S1, our skill capture glove, alongside our robot hand to record joint movements that map to robot targets, with touch and video on the same timeline.
1
3
18
998
Leena Mathur retweeted
Quanting Xie
@DanielXieeeOct 21
The glove and robot hand share the same joint arrangement. After calibration, each measured angle gives us a target for the corresponding robot joint. Watch the knuckle, middle and tip joints move independently. We want to preserve that detail in the recording.
2
7
31
3,445
Amazing to see what @DanielXieee and team have been building for the future of robot dexterity!
Quanting Xie
@DanielXieeeOct 21
A million hours of what? If we’re scaling human demonstrations for robot learning, what should each hour contain? We built S1, our skill capture glove, alongside our robot hand to record joint movements that map to robot targets, with touch and video on the same timeline.
1
8
525
Leena Mathur retweeted
Yonatan Bisk@ybiskOct 21
Gooooooood data
Quanting Xie
@DanielXieeeOct 21
A million hours of what? If we’re scaling human demonstrations for robot learning, what should each hour contain? We built S1, our skill capture glove, alongside our robot hand to record joint movements that map to robot targets, with touch and video on the same timeline.
2
17
1,080
Leena Mathur retweeted
Yonatan Bisk@ybiskOct 20
Discovering dexterity: At home with my infant, and at Origami @origami_robotic ... More tomorrow 👀
19
28
12
163
36,199
Leena Mathur retweeted
Emmy Liu@_emliuOct 20
On the job market this year! Open to faculty as well as RS/MTS positions and primarily interested in the science of language models. Please reach out if you are hiring and you think I would be a good fit.
5
51
285
18,032
Leena Mathur retweeted
Lindia Tjuatja@lltjuatjaOct 19
Some topics I'm looking forward to working on with my students below 🧵
Lindia Tjuatja@lltjuatjaOct 19
The UT comp ling faculty (@jessyjli, @kanishkamisra, @kmahowald & myself) are looking for PhD students to join us! sites.utexas.edu/compling/ 🤘 We are an interdisciplinary group with interests across linguistics, cogsci and NLP. Apply by 12/1: liberalarts.utexas.edu/lingu…
1
2
19
1,620
Leena Mathur retweeted
Lindia Tjuatja@lltjuatjaOct 19
The UT comp ling faculty (@jessyjli, @kanishkamisra, @kmahowald & myself) are looking for PhD students to join us! sites.utexas.edu/compling/ 🤘 We are an interdisciplinary group with interests across linguistics, cogsci and NLP. Apply by 12/1: liberalarts.utexas.edu/lingu…
1
33
7
136
19,111
Bridges in autumn. 🍂 🍁
6
67
5
395
6,326
Leena Mathur retweeted
Jeff Dean
@JeffDeanOct 16
Have a glimpse at this! Nice work, @AndrewDai and Elorian folks!
Andrew M. Dai
@AndrewDaiOct 16
Spent years watching models ace a single image and completely whiff on anything that moves or changes angle. Today we’re fixing that. Try Glimpse now, our first visual thinking model with competitive frontier performance compared to Astra on key public visual reasoning benchmarks.
43
51
6
828
105,900
Leena Mathur retweeted
Tesla
@TeslaOct 16
Roadster xitter.cf/i/broadcasts/1yKAPwNgg…
0:00

Tesla

Roadster

3,020
12,444
3,338
94,709
4,191,405
Multimodality + multilinguality!
Jake Boggs
@JakeABoggsOct 13
Gemini 4 Argon sets a new record of 69% on HieroglyphBench This is the first model that performs accurately enough to be useful
2
20
815
Leena Mathur retweeted
Yash Jangir
@off_jangirOct 13
Check out our latest work :) What 30,000 Hours of Egocentric Video Do Not Teach? It learned exactly where the hands go. What happens to the things they touch, much less so. fidelity-gap.github.io/ Amazing collaboration with @JiahuaDong @ptokmakov @Miccooper9 @mzubairirshad @ZakharovSergeyN @mangahomanga @vitorguizilini @ToyotaResearch

What 30,000 Hours of Ego-centric Video Does Not Teach

Scaling ego-centric human video from 300 to 30,000 hours — what improves with data, what saturates, and what does not.

fidelity-gap.github.io
Jiahua Dong
@JiahuaDongOct 12
[1/7] We trained a world model on 30,000 hours of first-person video. It learned exactly where the hands go. What happens to the things they touch, much less so. Going from 300 to 30,000 hours did not narrow the gap, and a 4× larger model barely did either.
6
15
880
Leena Mathur retweeted
Matthew Mason
@mastertoadsterOct 13
Will all robots someday be replaced by humanoids? No! Specialist robots will always be useful, even if we do someday achieve the dream of human-equivalent humanoid robots. mtmason.com/should-robots-be…

Should robots be generalists or specialists? - Matthew T. Mason

Will all robots someday be replaced by humanoids? No! Specialist robots will always be useful, even if we do achieve the dream of human-equivalent humanoid robots.

mtmason.com
1
1
4
273
If you’re applying to CMU’s School of Computer Science @SCSatCMU and are interested in the Language Technologies Institute’s PhD or MLT programs @LTIatCMU, come to our info-session on November 11! We’ll have faculty there to answer any questions you may have about the department and application process
Language Technologies Institute | @CarnegieMellon@LTIatCMUOct 13
The LTI is holding a virtual info-session for prospective MLT and PhD applicants on Wednesday, November 11 from 12pm-1pm, ET! Our faculty will be answering any questions about the application process. Register at the following link to attend: forms.gle/pxsdNCY4JXZmjUxJ6
1
2
9
678
The LTI is hiring for teaching track and tenure track faculty positions to start in the Fall 2027 semester. See our Careers page for full details and access to the application portal. Note: Teaching Track applicants are encouraged to apply by October 23! lti.cmu.edu/careers/index.ht…

Careers - Language Technologies Institute - Kenneth C. Griffin School of Computer Science -...

Career Opportunities at Carnegie Mellon University's Language Technologies Institute

lti.cmu.edu
8
1
33
10,595
Leena Mathur retweeted
Katie Jacobs Stanton
@KatieSOct 10
🇺🇸 Watching the @BlueAngels is equal parts awe, inspiration and gratitude.
2
30
255
8,775
Leena Mathur retweeted
CLS
@ChengleiSiOct 10
I had the fortune of listening to Jeremy Avigad discussing the future of mathematics in the age of AI, and I find his following slides deeply inspiring
7
3
49
3,658
Leena Mathur retweeted
Daniel Fried
@dan_friedOct 9
Zora is presenting this (and her other awesome work) as an invited talk at LSEI now - Union Square 15&16! #COLM2026
Zora Wang@ZhiruoWSep 8
Agents trained on population-scale data can do a lot of things. But ask a professional to stake their reputation on an AI-generated artifact? “Pretty good” isn’t good enough. We introduce TAHI: a Test-time Adaptive agent framework through Human-agent Interaction. Featuring: 📈 Test-time adaptation via context (memory, skills) and weight training ⚡ Efficient adaptation to individual expertise within tens of tasks ✔️ Creating comprehensive rubrics for “non-verifiable” tasks 🔍 Analysis of shared community guidelines vs. personalized tacit expertise
3
35
3,163
Load more