David Mueller

Presenting my poster on dataset documentation today at 5pm (#619)! If you liked Jelani Nelson's keynote yesterday on sketching, come see how we use efficient datastructures to document datasets at crazy compression: 800 GB dataset ➡️ 27 GB 🎉🎉🎉

I have written a blogpost on balancing multiple directions in multilingual translation, which covers some of the latest advances in #NeurIPS2023. Link: tianjianl.github.io/blog/2023/mnmt/

Can I trust this AI/ML prediction? A good uncertainty quantification (UQ) method could answer this with a predictive confidence interval. But in practice, 2 key challenges limit real-world UQ. Our #ICML2023 Oral “JAWS-X” tackles these challenges! 🧵w/ @suchisaria @anqi_liu33 1/n

The competition gets stiffer every year

Call for reviewers for our #ICLR2023 workshop on Mathematical and Empirical Understanding of Foundation Models. Fill out this form if you are interested sites.google.com/view/me-fomo20… and we will aim to get back to you asap. Paper deadline: 3 Feb Tentative reviewing period: 10-24 Feb.

Introducing task vectors! A new way to steer models by doing arithmetic with model weights. Subtract to make models forget, add to make them learn 📜: arxiv.org/abs/2212.04089 🖥️: github.com/mlfoundations/…

I want to talk about our #EMNLP2022 paper, led by @dam_nlp with Nick Andrews. Do Text-to-Text Multi-Task Learners Suffer from Task Conflict? cs.jhu.edu/~mdredze/publi… We've got some pretty interesting analyses for those working in multi-task language model training.

High-Dimensional Probability math.uci.edu/~rvershyn/teac…

Our #EMNLP2022 paper compares the effectiveness of three span finding methods for information extraction tasks: Tagging, Span Enumeration, and Boundary Prediction. See the full paper here: arxiv.org/pdf/2210.06824 (w/ Weiwei Gu, @Boyuan62136063, @ctongfei, @ben_vandurme)🧵1/6

Come see our presentations at EMNLP! taur.cs.utexas.edu/emnlp2022.shtml All of these papers represent directions in textual reasoning and summarization that have been in the works for a while and I'm excited that we're finally sharing them! Up tomorrow: @tanyaagoyal and @ZayneSprague

Given I am lucky enough to be at #NeurIPS2022 in person, I am proud to present our poster on collecting and learning from ambiguous images and human uncertainties this afternoon Hall J #1002. This is a joint work with amazing collaborators @kesnet50 @ben_vandurme and Reno Kriz.

@EliasEskin Semantic Santa, is that you?

🎁 Early Holiday Preprint 🎄 Task-oriented semantic parsing is used in interactive systems, where calibration really matters. We find they're surprisingly well-calibrated and use that to build DidYouMean, a system for confirming user intent arxiv.org/abs/2211.07443 🧵 1/12

#NeurIPS2022 group run 8am starting from Lafayette Square tomorrow (Thursday), everyone is welcome! We plan to run north along the river (passing through French Quarter)

@GRESEARCHjobs Thank u 🤗

@dam_nlp We do! Cathy is serving delicious coffees on our booth (203) all day. See you there! ☕️

Important #NeurIPS2022 question: does @GRESEARCHjobs have a barista, and if not where do I find yummy coffee?

If you’re at #NeurIPS2022 come see our poster Hall J #941 TODAY (11/29) 4-6pm! With @anqi_liu33 @suchisaria, we present “JAWS: Auditing Predictive Uncertainty Under Covariate Shift” Virtual talk: neurips.cc/virtual/2022/p… Paper: arxiv.org/abs/2207.10716…

What is a work boat?

Yes! Come check my poster during session 2 @SymposiumML4H (4:40pm)

@rdesh26 Optimistically or realistically? ;)

Getting on a 15.5 hour ORD-DEL flight tonight. How will I spend this time?

Is there anyone attending NeurIPS who is a leader at a funding body such as NSF, CIFAR or [[insert here other national or supranational funding body]]? I'm moderating a panel for NeurIPS on "How has funding and incentive structures impacted the history of computer science?"