"in 2025 we will have flying cars" 😂😂😂

🚨New at #ICLR: we introduce the first ever 𝐥𝐚𝐲𝐞𝐫 that makes 𝐚𝐧𝐲 neural network 𝐜𝐨𝐦𝐩𝐥𝐢𝐚𝐧𝐭 𝐛𝐲 𝐝𝐞𝐬𝐢𝐠𝐧 with constraints expressed as 𝐝𝐢𝐬𝐣𝐮𝐧𝐜𝐭𝐢𝐨𝐧𝐬 𝐨𝐟 𝐥𝐢𝐧𝐞𝐚𝐫 𝐢𝐧𝐞𝐪𝐮𝐚𝐥𝐢𝐭𝐢𝐞𝐬—even if they define 𝐧𝐨𝐧-𝐜𝐨𝐧𝐯𝐞𝐱 𝐬𝐩𝐚𝐜𝐞𝐬!

RT @BohuslavskaKate: At least 24 people russia murdered in Sumy today. 24 people gone, just like that. At least 84 got injured, includi…

Green Day calls out JD Vance during "Jesus of Suburbia" theneedledrop.com/news/green…

Volodymyr Zelensky is NOT a dictator. Vladimir Putin IS a dictator. Zelensky didn’t start the war in Ukraine. Putin did. Hope this clarifies things for anyone who is confused.

Finally, in frustration, I asked if it could write a regex to find all occurrence of the bracketed numbers in the output. It got the regex wrong. What's going on here? I feel like the emperor has no clothes. Are people solely using these tools to write boilerplate? (10/10)

Hiring a postdoc/RA as part of my UKRI Future Leaders Fellowship, with flexibility and the chance to develop an independent research portfolio. ➡️ Length: 15 months starting from April 2025 or soon after ⏰ Application deadline: 3 January ✍️ How to apply: krb-sjobs.brassring.com/TGne…

New fully funded PhD scholarship at @Cardiff_NLP: 🖥️ Topic: Flexible as long as related to NLP 📅 Application deadline: January 3rd 🏃Start date: Flexible (April/July/October 2025) Please share and contact me if you have any questions! findaphd.com/phds/project/ph…

I hope I'm wrong, and that there's something I'm not seeing, but as far as I can tell, this appears to be a craven and somewhat pathetic bid from the Nobel committee to capitalize on AI hype by arbitrarily changing their standards, and I find it all rather distasteful. (6/6)

Me when Black Myth Wukong drops

Even though it’s rather outside my area of expertise, «On the Independence Assumption in Neurosymbolic Learning» by @EmilevanKrieken and @PMinervini communicates and leaves the proofs in the paper. #ICML2024

❤️

[CL] The Geometry of Categorical and Hierarchical Concepts in Large Language Models K Park, Y J Choe, Y Jiang, V Veitch [University of Chicago] (2024) arxiv.org/abs/2406.01506 - The paper studies how categorical concepts like {mammal, bird, reptile, fish} and hierarchical concepts like animal vs mammal are represented in large language models. - It extends the linear representation hypothesis, which says high-level concepts are linearly encoded, from binary concepts to categorical and hierarchical concepts. - It shows categorical concepts are represented as simplices, with vertices corresponding to binary features like mammal, bird, etc. - It shows hierarchical concepts are encoded through orthogonality. For example, the representations of animal and mammal are orthogonal. - It validates these theoretical results empirically on the Gemma large language model, estimating representations for 957 hierarchically related concepts from WordNet.

CFP: The #IJCAIworkshop on Logical Foundations of Neuro-Symbolic AI at IJCAI'24 is calling for papers. Link: sites.google.com/view/lnsai2… Deadline: May 8.

The 3rd Cardiff NLP Workshop is just around the corner, join us! 🔜

Kharkiv, Ukraine's 2nd largest city, is under relentless missile and Shahed drone attacks, mercilessly targeting innocent civilians. In just the past weeks, these barbaric assaults have taken the lives of at least 15 people, leaving countless others wounded, including vulnerable children. Ukraine urgently requires additional air defense systems to shield its citizens. Every second counts. #UkraineNeedsAirDefense

The 3nd Cardiff NLP Workshop is around the corner! 📅 1-2 July 2024 in Cardiff (in-person only) 🎇 All about #NLProc: invited talks, tutorials, panels, posters, networking 🧑‍🤝‍🧑 Join us by completing the expression of interest form by April 8th!

Unveiling the next chapter in our quest towards the elusive theory of *all* neural networks: Categorical Deep Learning! 🐈🤖 categoricaldeeplearning.com See thread 🪡 for a bucket list 🪣 of concepts our framework can handle! w/ @bgavran3 P. Lessard @andrewdudzik @tlvg @_joaogui1

When you’re not storing any text; you’re just storing weights and generating things