Some messages from the DMLR data curation workshop: Lucas Beyer (bl16) proposed an experiment for measuring performance when emergence may happen-- please write a tweet about it ! Synthetic data works for text but not for so much for images right now. We think it's because of spurious

Really Enjoyed the Data-Centric Machine Learning workshop in ICML🤠

This is the most scary Halloween costume I've seen 👻

Precisely correct. SB 1047 is an attack on AI research itself and will hurt both the creation and distribution of models, particularly open weight models.

Thank you Fei-Fei

Huge! Multiple democratic California members of congress have submitted a letter to Gavin Newsom against SB 1047! Signatories including Rep. Zoe Lofgren, Rep. Anna G. Eshoo, Ro Khanna, Rep. Scott Peters, Rep. Tony Cárdenas, Ami Bera, M.D., Nanette Barragán and Rep. Lou Correa Please help amplify!

Interesting question, let us know.

Does style matter over substance in Arena? Can models "game" human preference through lengthy and well-formatted responses? Today, we're launching style control in our regression model for Chatbot Arena — our first step in separating the impact of style from substance in

Hey collective Twitter AI hivemind: which Supervised FineTuning library should we use for our research? We are exploring libraries for SFT, e.g. TRL, Axolotl, Torchtune, or other options?

I just discovered the Error Correcting Zoo: A very cool resource for Coding theory

GPT is having a profound effect on how students write. Its verbose style, full of cliches and 'fancy', out of place vocabulary is in every paper and draft I read. A few years back, there were grammar errors and awkwardness -- but at least people had their own voice. Now,