Dynamic Adversarial Data Collection Workshop retweeted
I really enjoyed my @MLStreetTalk chat with Tim at #NeurIPS2024 about some of the research we've been doing on reasoning, robustness and human feedback. If you have an hour to spare and are interested in some semi-coherent thoughts revolving around AI robustness, it may be worth a listen 🎧
Check it out at youtu.be/DL7qwmWWk88?si=bCvt…
Dynamic Adversarial Data Collection Workshop retweeted
We are honoured to have the amazing @VeredShwartz giving a talk at our UCL NLP Meetup titled "Incorporating Commonsense Reasoning into NLP Models"! Join us later today, Friday 29th of July, at 4:30pm BST: meetup.com/ucl-natural-langu…
Dynamic Adversarial Data Collection Workshop retweeted
New blog post 📚: Insights from #NAACL2022 on #multimodality, #explainability, data collection & #benchmarking, #bias, and some further papers I enjoyed reading!
mubasharaakhtar.com/post/naa…
Dynamic Adversarial Data Collection Workshop retweeted
The First Workshop on Dynamic Adversarial Data Collection (DADC) @DADCworkshop , co-located with NAACL 2022: (lnkd.in/ghvAmJJd)
This paper lnkd.in/gtwfQBrE analyses the Dynamic Adversarial Data Collection protocol on the annotator’s experience and the data quality.
Dynamic Adversarial Data Collection Workshop retweeted
Here's a recording of my talk (no NAACL registration necessary):
youtu.be/JcNpiD4odT0
We have an incredible panel lined up for you tomorrow on the Future of Data Collection! Moderated by @adinamwilliams, panelists include @annargrs, @boydgraber, @sleepinyourhat, @tongshuangwu, @laroyo, @douwekiela & @swabhz 🤩 Post any questions you have for them below! 🚀
Dynamic Adversarial Data Collection Workshop retweeted
I know I am biased, but how cool are the @DADCworkshop stickers??🤩 Come join us in room Duc Sol to claim yours, as we are still here with presentations of our winners on the Shared Task!🙌🏻 #DADC22 #NAACL2022
Dynamic Adversarial Data Collection Workshop retweeted
Okay #NAACL people....Get ready 🏁 Only 20 minutes to go until our flagship event 😬🍿 this panel is the stuff dreams are made of for any NLP enthusiast ☁️🦄💥 Don't miss out on some wholesome heated debate and join online or in room 708. We're ready for some serious 🔥 #DADC22
We have an incredible panel lined up for you tomorrow on the Future of Data Collection! Moderated by @adinamwilliams, panelists include @annargrs, @boydgraber, @sleepinyourhat, @tongshuangwu, @laroyo, @douwekiela & @swabhz 🤩 Post any questions you have for them below! 🚀
Dynamic Adversarial Data Collection Workshop retweeted
Slides here!
cims.nyu.edu/~sbowman/dadc_s…
We're delighted to announce @sleepinyourhat (@NYUniversity / @AnthropicAI) as a keynote speaker discussing "Why Adversarially-Collected Test Sets Don't Work as Benchmarks". More fantastic speakers to be announced soon, stay tuned! 🚀
Dynamic Adversarial Data Collection Workshop retweeted
This was totally 100% intentional and part of my longstanding commitment to high-quality hybrid interactions and not the random result of my arbitrary green screen / monitor setup.
Our audience interaction @naaclmeeting transcends online-offline worlds 🤯 Who needs a metaverse when you have @boydgraber ? 😅 #NAACL #DADC22
Dynamic Adversarial Data Collection Workshop retweeted
I'm starting in a couple of minutes. If you're at #naacl2022 and interesting in adversarial data collection or grumpy talks, stop by!
We're delighted to announce @sleepinyourhat (@NYUniversity / @AnthropicAI) as a keynote speaker discussing "Why Adversarially-Collected Test Sets Don't Work as Benchmarks". More fantastic speakers to be announced soon, stay tuned! 🚀
Dynamic Adversarial Data Collection Workshop retweeted
Does adversarial data collection solve our benchmark problems? Really looking forward to @DADCworkshop panel & invited talks!
Panel 13:55 Seattle/22:55 CEST
@annargrs 9:10/18:10
@boydgraber 9:45/18:45
@sleepinyourhat 11:10/20:10
@tongshuangwu 11:45/20:45
@laroyo 13:20/22:20
We have an incredible panel lined up for you tomorrow on the Future of Data Collection! Moderated by @adinamwilliams, panelists include @annargrs, @boydgraber, @sleepinyourhat, @tongshuangwu, @laroyo, @douwekiela & @swabhz 🤩 Post any questions you have for them below! 🚀
Dynamic Adversarial Data Collection Workshop retweeted
Peter Mattson (@MLCommons, Google) at @DADCworkshop:
> Data is the new code, model is the compiler.
The talk presented dataperf.org/, a very cool initiative. We have tons of fixed-benchmark leaderboards; how about fixing the model & trying to come up with better data?
Dynamic Adversarial Data Collection Workshop retweeted
I will present this work orally at 14:15 in Columbia D on Wednesday and as a poster at 15:40 @DADCworkshop on Thursday (1/3) nitter.cf/ruixiangcui/status/151…
Excited to announce that our work “Generalized Quantifiers as a Source of Error in Multilingual NLU Benchmarks” has been accepted at #NAACL2022! With @daniel_hers and Anders Søgaard @coastalcph, we investigate the presence of generalized quantifiers in NLI/QA tasks. 1/4
Dynamic Adversarial Data Collection Workshop retweeted
I'm excited to attend my first in-person *ACL event #NAACL2022 in my home city of Seattle! I'm happy to chat about question answering, evaluation/benchmarking, and dataset construction! Old friends, reach out! Also, be sure to come to our @DADCworkshop on Thursday!
Dynamic Adversarial Data Collection Workshop retweeted
Preprint: "How many linguists does it take to fool a question answering model?"
We take a systematic linguistically informed approach to adversarial attacks, obtaining over 60% Model Error Rate and an getting (unverified) #1 in @DADCworkshop shared task.
arxiv.org/abs/2206.14729
Dynamic Adversarial Data Collection Workshop retweeted
Excited to see this work by @max_nlp and colleagues. on how generative models can help data annotation.
If you've never heard about Dynamic Adversarial Data Collection, the #NAACL DADC Workshop website covers the basics and links to the literature:
dadcworkshop.github.io/share…
Data collection is slow and expensive, so we give annotators a little help 🤝. Introducing Generative Annotation Assistants (GAAs) to make data collection more efficient and effective 🚀. Work will be presented at #NAACL2022 in Seattle!
Paper: arxiv.org/abs/2112.09062
[1/n]
Dynamic Adversarial Data Collection Workshop retweeted
🚨 New paper and datasets! 🚨
After sitting on my hands for many months 😬 I'm delighted that our #Hatemoji paper is going to @naaclmeeting! 😍🤩😎🆒
In a nutshell 🥜it uses human-and-model-in-the-loop learning 🤖🤝🙆 to tackle emoji-based hate
A 🧵 on all our new resources 1/
Dynamic Adversarial Data Collection Workshop retweeted
"@DynabenchAI relies on crowdworkers"...
But it doesn't have to! The @DADCworkshop shared task is a fantastic opportunity for the wider #NLProc community to contribute -- and it's currently underway. Can YOU beat the AI?: dynabench.org/tasks/qa 🚀
science.org/content/article/…
Dynamic Adversarial Data Collection Workshop retweeted
The Multilingual Spoken Words dataset from MLCommons is on the @huggingface Hub 🤗🤗🤗
huggingface.co/datasets/MLCo…
It's a large audio dataset of spoken words in 50 languages, including low-resourced ones, for keyword spotting and spoken term search.

























