🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Excited to present our paper on Semantic Uncertainty at ICLR in Kigali this week! Join us at 3:40 pm in AD11 on Wednesday, as part of Oral 6 Track 2. Also, do get in touch if you'd like to talk about uncertainty, interpretability or robustness in LLMs!

With more powerful AI systems comes more responsibility to identify novel capabilities in models. 🔍 Our new research looks at evaluating future 𝘦𝘹𝘵𝘳𝘦𝘮𝘦 risks, which may cause harm through misuse or misalignment. Here’s a snapshot of the work. 🧵 dpmd.ai/novel-ai-risks

The Google DeepMind alignment team is looking for research scientists and research engineers to help us work towards safe AGI. I think this is a very pressing problem, and it's a nice place to work. Please apply and help take our work to the next level. boards.greenhouse.io/deepmind/jobs/…

new preprint "ReLU to the Rescue: Improve your On-policy Actor-Critic with Positive Advantages" shockingly simple changes to A3C can give a cautious RL algorithm more effective than PPO in some settings, just adding a ReLU is enough! arxiv.org/abs/2306.01460

Excellent post by Jacob Steinhardt trying to forecast the abilities of models that could be trained in 2030: bounded-regret.ghost.io/what-will-gpt-…

rainy day in sf

But the ELO can ultimately become bounded by the difficulty of the prompts (i.e. can’t achieve arbitrarily high win rates on the prompt: “what’s up”). We find on harder prompt sets — and in particular coding — there is an even larger gap: GPT-4o achieves a +100 ELO over our prior

Today, I’m excited to share with you all the fruit of our effort at OpenAI to create AI models capable of truly general reasoning: OpenAI's new o1 model series! (aka 🍓) Let me explain 🧵 1/

We trained a model and it is good in some things

very excited about these models helping people solve hard problems and proud of the work we did. give the new models a try!

i generally feel super grateful that i get to work with such exceptionally skilled and kind people on reasoning research. the sprint for IOI in particular was special though. IOI 2024 gold @ 10k submissions; 49th percentile of competitors under real contest conditions

Two important points from our new technical report: 1. Scaling continues to work and the bitter lesson still holds 2. Recent AI models are strong at reasoning tasks and are rapidly becoming stronger — 4o was released less than a year ago, o1 less than six months ago

Congratulations Psyho on a nail-biting performance! Great showings as well from Borys Minaiev, Andre Saraiva, and Lorenz Kuhn representing OpenAI. It’s been fantastic sponsoring AtCoder World Finals AtCoder. We’re excited to share some of the model solutions with the world.

It was thrilling to watch AI compete against some of the best human competitive programmers at AtCoder World Finals Heuristics yesterday. Check out Andre Saraiva ‘s thread on how the AI solutions improved throughout the 10h contest. Congrats to Psyho on 1st place!

IMO gold is a win for scaling ~nearly~ superhuman oversight on a fuzzy, hard-to-verify RL domain