Language is hierarchical. So, shouldn’t our coreference resolution models be as well? Yes! And, on that note, we are excited to share our #NAACL2022 paper “A Structured Span Selector”, which discusses our grammar-based, neural coreference resolver. #TeamStructure
I’m at #NAACL2022 in Seattle! There is a donut wall at the conference! Come talk to me about grounded language and donuts.
Hi #NAACL2022 ! Last summer we had a crazy idea of distilling transformer models into shallow, sparse, and fast models. Curious about whether and to what extent this idea works? Please come to our presentation tomorrow!
📍 Session 1D @ Elwha A
⏰ Mon 11:30-11:45
Introducing Fig-QA #NAACL2022 !
arxiv.org/abs/2204.12632
Can models such as BERT, RoBERTa, and GPT-* reason about figurative and metaphorical language well, especially when said language is creative or unusual? (1/11)
Enjoying #NAACL2022 and Seattle? The party is not over yet! The workshop on Structured and Unstructured Knowledge Integration (SUKI) is happening on Thursday! Come to join our amazing lineup of speakers and discussions. Details on website: suki-workshop.github.io
I’ll present our paper “On the Robustness of Reading Comprehension Models to Entity Renaming” at #NAACL2022 . We study the generalization ability of MRC models to novel entity names, by evaluating their robustness to entity renaming.
📅Tue(07/12) 16:30-16:45
📍Session 6E@Elwha B
Excited that Spider will be presented orally at #NAACL2022 🚀
1. Code is now available!
2. All models are at Hugging Face hub (plus *a new model* - reached 84% recall@100 on NQ without any supervision)
3. Finally a dense model to pass BM25 on the challenging EntityQuestions!
This is one of the most inspiring slides I encountered at #NAACL2022 . Big Shout out to hci.social/@jbigham - can I quote it lol?
When NLI models can ignore context, do they? Our #NAACL2022 paper explores this + revisits the role of partial-input baselines in NLI! Joint work w/ Rachel Rudinger
🔗 arxiv.org/abs/2205.12181
I'll be presenting this today:
📆 In-Person PS 3 (10:45 PT)
📆 Virtual PS 2 (14:15 PT)
It is great to see about 200 audiences attending! Come join us this afternoon at the Structured and Unstructured Knowledge Integration (SUKI) workshop #NAACL2022 !
🤔 How can we systematically measure whether large language models (LLMs) encode human stereotypes?
💡 Let's lean on well-established models of human stereotyping from social psychology!
🗓️ #NAACL2022 session 5A, Tue 12 Jul, 14:15 PT
📄 arxiv.org/abs/2206.11684
Presenting our work “On Curriculum Learning for Commonsense Reasoning” at #NAACL2022 poster session PS2-5 (in-person) today & PS1-9 (virtual) tomorrow!
Paper: aclanthology.org/2022.naacl-mai…
Code: github.com/adymaharana/cu…
Underline: underline.io/events/325/pos…
cc Mohit Bansal (@uncnlp)
🧵
Now that #NAACL2022 is over, I just want to share this rare find of a cup of coffee left unattended on the counter in a coffee shop nearby the venue. Apparently it belongs to Christopher Manning, but I have no idea why a simple cup costs 30+ dollars and why it was there 🤣
Multi-task NLP systems such as T0 can tackle *novel tasks* by leveraging prior task knowledge. While this transfer is interesting, we test whether biases picked up during training leak into a novel task.
New paper to appear in findings of #NAACL2022 !
arxiv.org/pdf/2205.11605…
I’m in Seattle #NAACL2022 ! My first time in the US and my first physical conference ever! Let’s meet and chat about multilinguality, logic reasoning, semantic parsing, etc.