ICML 2024: Best Paper Awards: Florian Tramèr; Gautam Kamath; Nicholas Carlini. Position: Considerations for Differentially Private Learning with Large-Scale Public Pretraining Akbir Khan; John Hughes; Dan Valentine; Laura Ruis; Kshitij Sachan; Ansh Radhakrishnan; Edward

Test-of-time Award ICML 2024. DeCAF: a deep convolutional activation feature for generic visual recognition Jeff Donahue, Yangqing Jia, Oriol Vinyals, Judy Hoffman, Ning Zhang, Eric Tzeng, and Trevor Darrell

With today’s launch of our Llama 3.1 collection of models we’re making history with the largest and most capable open source AI model ever released. 128K context length, multilingual support, and new safety tools. Download 405B and our improved 8B & 70B here.

Llama 3.1 sets a new standard for open source AI. The 8B, 70B, and 405B delivers performance comparable to the most advanced closed models. Here’s a look at how the performance stacks up. You can dive deeper in the full announcement and research paper available here:

A few people are asking me why work with AI at Meta? Two words: Open Source.

Machine learning models must be audited for fairness, robustness and accuracy, yet many are confidential. How can we audit a confidential model for fairness while still retaining confidentiality? We show how to do this in our #ICML2024 paper.

Zeyuan Allen-Zhu Ok, I understand the issue now. We were able to restore the video for now, but the video you posted used a combination of your own recording plus the SlidesLive stream from ICML. The ICML contracts require that direct video recordings from the SlidesLive stream not be publicly

Please see an explanation by Zico Kolter as to why ICML tutorial video was initially taken down and is now restored.

CMU-MATH Team’s Innovative Approach Secures 2nd Place at the AIMO Prize – Machine Learning Blog | ML@CMU | Carnegie Mellon University blog.ml.cmu.edu/2024/07/29/cmu…

"Proud Advisor Moment": Congratulations to Paul Liang Paul Liang on joining MIT as an assistant professor MIT EECS and MIT Media Lab. If you are interested in working on the next generation Multimodal Multisensory AI, I would strongly consider joining Paul’s new lab at MIT.

From the organizing committee and everyone involved we hope that you had an amazing time at ICML

We’re opening a call for applications for the next round of Llama Impact Grants. The program supports organizations pursuing ideas to use Llama to address social challenges in their communities with $2M in grants available. Full details + criteria: ai.meta.com/blog/llama-3-1…

Unfortunately, I couldn't make it to ACL in person this year, but I'll be presenting VisualWebArena at virtual poster session #2 next Thurs: ⏰ August 22, 10.15am-11am (8.15pm-9pm PST) 📝 arxiv.org/abs/2401.13649 🌐 jykoh.com/vwa Come and chat about multimodal agents!

Causal Inference under Incentives: An Annotated Reading List – Machine Learning Blog | ML@CMU | Carnegie Mellon University blog.ml.cmu.edu/2024/08/22/cau…

It is exciting to see CMU School of Computer Science CMU School of Computer Science highlighting #ECCV2024 work of So Yeon (Tiffany) Min on Situated Instruction Following -- Teaching AI Agents To Navigate Ambiguous Instructions: cs.cmu.edu/news/2024/robo… This work generalizes Embodied Instruction

I recently moved to the Bay Area! Excited to work with Russ Salakhutdinov and the team AI at Meta on agents 🦙

Welcome Shuyan Zhou! Looking forward to working with you!