You heard it right: I obsessively took over 15k pictures over a multiyear period, worked with an amazing team to get them cleaned up, vetted, and captioned, and donated them for research. Hoping others will donate images in similar manner in future!

Excited to introduce GPT-4o. Language, vision, and sound -- all together and all in real time. This thing has been so much fun to work on. It's been even more fun to play with -- with moments of magic where things feel totally fluid and I forget I'm video chatting with an AI.

👐 I'm looking for a couple of emergency reviewers who can help out with a paper for COLM in the area of agents. Please reply or DM me for details. If you finished Neurips submissions and have some free time to help out, your contribution is much much appreciated!!! 😃 🦾

How well do DPO and PPO work on public preference datasets? Excited to share some work exploring the effects of data, reward models, and prompts! We also find that PPO generally beats DPO, despite being more challenging engineering-wise. 📜: arxiv.org/abs/2406.09279 More below 👇

When talking abt personal data people share w/ OpenAI & privacy implications, I get the 'come on! people don't share that w/ ChatGPT!🫷' In our Conference on Language Modeling paper, we study disclosures, and find many concerning⚠️ cases of sensitive information sharing: tinyurl.com/ChatGPT-person…

The Art of Saying No Contextual Noncompliance in Language Models Chat-based language models are designed to be helpful, yet they should not comply with every user request. While most existing work primarily focuses on refusal of "unsafe" queries, we posit that the scope of

We've publicly released our DataComp-LM models: Truly open 1B and 7B models that's competitive with state-of-the-art (llama3, qwen2, gemma, ...) on most benchmarks, but with a public training recipe, dataset, and code! (1/3)

New short talk on the paper: Unpacking DPO and PPO -- Disentangling Best Practices for Learning from Preference Feedback (Hamish Ivison et al) I go through our thought process and experimentation of trying to show that PPO >> DPO with open datasets and models. Builds on the Tulu

Interested in working on Gemini pre-training? I'm hiring a research scientist to work on pre-training data Google DeepMind in London: boards.greenhouse.io/deepmind/jobs/… I am unfortunately not at #NeurIPS2024 but feel free to reach out to ask questions or see the team at the booth there!

🚨Test data is out! 🚨 The testing phase will run until May 24, 5 pm PT. Check out our github for the data + submission instructions. Bring your best models 💪! Participants can also submit shared task reports to Scholarly Document Processing Workshop after the testing phase!

Life update: I’m excited to share that I’ll be starting as faculty at the Max Planck Institute for Software Systems(Max Planck Institute for Software Systems) this Fall!🎉 I’ll be recruiting PhD students in the upcoming cycle, as well as research interns throughout the year: lasharavichander.github.io/contact.html

Attending #ACL2025NLP in Vienna this week! If you're around and want to chat about multimodality, audio, vision, or reasoning, let’s connect! Check out some of our recent work from Mistral: 🔊Voxtral: arxiv.org/pdf/2507.13264 🧠 Magistral: arxiv.org/pdf/2506.10910