Excited to co-organize Table Representation Learning workshop at ACL this year! Please submit your work about Language + table/structured data. Submission deadline: ⭐️April 15th⭐️

Somehow every time I see animals solving a problem, it reminds me of @ylecun.bsky.social perspective. Two questions: Can a reasoning model (e.g. O3) solve it within a sandbox (I think it is harder than ARC-AGI)? And can multiple agents or robots collaborate to tackle the problem together?”

Deliberative alignment: reasoning enables safer language models Introducing our new alignment strategy for o1 models, which are directly taught safety specifications and how to reason over them.

I’ve recently been thinking about the limits of RL only for alignment and the potential of self-reflection reasoning for safety. It’s nice to see OpenAI releasing a report finally with a METHOD section leveraging both SFT and RL for alignment.

openai.com/index/delibe...

📣 Tomorrow (Saturday) #NeurIPS2024 , 8:40am 😱
Table Foundation Models:
How we finally got deep neural networks to markedly outperform gradient-boosted trees ✨✌️
table-representation-learning.github.io

So sorry we didn’t get a chance to say hi but it was a pleasure introducing our work to you. Thank you for your interest in our work!

looks so pretty! Will check it out

Where is it!

Let’s catch up tomorrow!

On my way to Vancouver! I’ll be presenting our RAGChecker poster tomorrow afternoon on behalf of my amazing Amazon colleagues. If you’re interested in RAG, especially RAG Evaluation, please check out our poster and paper!
𝐖𝐞𝐝 𝟒:𝟑𝟎-𝟕:𝟑𝟎 𝐖𝐞𝐬𝐭 𝐁𝐚𝐥𝐥𝐫𝐨𝐨𝐦 𝐀-𝐃 𝐏𝐨𝐬𝐭𝐞𝐫 𝟓𝟐𝟏𝟏
arxiv.org/abs/2408.08067