Hardy

the OpenAI goblin fiasco was a Big L for the interpretability research community They solved the mystery without SAEs or probing or anything. just talked to various models and counted the number of times they said Goblin

@vlachos_nlp Congrats!

Julius did amazing work! Very happy about this award, thank you!

Bing chat is so good that the Edge browser is soon to become my main browser replacing Chrome

🚨 Exciting news! We are delighted to announce NusaCrowd, a new open-source initiative to collect and unite Indonesian NLP resources! 🇮🇩🇮🇩🇮🇩 Through NusaCrowd, we have gathered 137 datasets and 117 standardized data loaders covering text, audio, and image modalities💪🏾✨💕🌍

After a year of restless development, I'm finally happy to announce Inseq, a new tool to democratize post-hoc interpretability of sequence generation models 🐛 github.com/inseq-team/ins… #nlproc #xai Some highlights 👇 1/

@fblain @TilburgU @TilburgU_DCA Congratulations Fred ! 🎉

Today I am joining @TilburgU & @TilburgU_DCA as Assistant Professor in AI. Can't wait to see what this next chapter entails! Stay tune and get in touch if you are interested in collaborating on topics such as MT, QE, PE and explore CogScience for Human Interpreting.

Fully Funded (including for International Students) PhD studentships are available for work on explainable #NLProc and #socialmedia. Offers go out starting Jan Work with me @UniofBath on Language Models and explainability. More at: harishtayyarmadabushi.com/research/four-… @ARTAIBath #phd 1/

JOB AD: We have a 3 year postdoc position on our fact-checking ERC grant AVERITEC starting in January: jobs.cam.ac.uk/job/38085/ or as soon as possible after that! If you know anyone who might be interested, send them my way! I will be at EMNLP for a chat! Deadline Dec. 15!

@FajriKoto Agree, thunderbird's search is the best

Why is Outlook Search really pathetic (and frustrating)?

Has anyone tried the Unhook chrome extension for Youtube? It's a must for digital minimalism.

@yanaiela It started as a science insight and turned into an engineering insight at some point which is different in everyone's opinion

And if we're talking about work that "stands the test of time", which part of these works would it be? that larger=better? It's an engineering insight, not science.

I don't really understand this claim. Sure, if you work at one of 5 companies that actually have the compute to run these huge models, you can just go larger. But even then, can you actually use these huge models in production? Are the improved results actually worth the money?

Kim Jung Gi was so phenomenal it was hard to believe anyone could do the things he did. Completely unreal he died so suddenly. He was only slightly older than me.

So the North American job market is starting to open up so here to say: I am on the market this year. My expertise is in content moderation (technical and theoretical), ethics of machine learning and ML’s philosophical inheritance. Core to my work is a search for equitable ML.

@kmkurn Serasa banget dimanjakan oleh sederhananya bhs indo, beneran terasa belajar bahasa itu sulit ketika mulai belajar bhs prancis yang ada gender dsb

🚨 Excited to share our new multilingual dataset, NusaX, covering 10 linguistically diverse low-resource Indonesian 🇮🇩 languages on the sentiment analysis & MT tasks. 📰 arxiv.org/pdf/2205.15960… We released the dataset github.com/IndoNLP/nusax #indonlp #nusax #NLProc @aclmeeting

The ProbAI 2022 materials 📚 and recorded lectures 🎥 are all available to the public. github.com/probabilistica…

@harish @UniofBath Congratulations!

Delighted to be starting as a Lecturer (~Assistant Professor) in Artificial Intelligence @UniofBath! Focus on #NLProc & #AI. Get in touch if you are interested in a PhD in #NLProc to do your PhD at a beautiful and leading UK university. View out of my office window:

Interested in visiting South Korea for an #NLP or #ML research internship? @kaist_ai is recruiting visiting researchers and interns at all skill levels (undergrad -> PhD) for this winter. Deadlines soon! gsai.kaist.ac.kr/winter-2022-ka…

@JunjieHu12 All my reviewers gave the same number for both scores. I think most reviewers are confused too.

Question for the EMNLP'22 review: there're two recommendation scores on the long paper submissions. what does that mean? Can a long paper submission be accepted as a short paper if it can be converted to 5 pages in the camera-ready version?

@colinraffel @srush_nlp This is an interesting approach on teaching!