Gene Li (@geneli0) profile picture

https://teapawt.org/t.co/5Bpr5cdNBE

Joined February 2017
Even with realizability we cannot avoid exponential sample complexity - but the precise answer turns out to depend on how we measure “quality” of the reset.
1
1
845
The lower bound constructions here were obtained via discussion with GPT 5.6. They are fairly clean and capture the intuitions that I had arrived at when I (unsuccessfully) worked on this problem in my PhD, e.g. the poisoning lock in Thm 1 and recursive amplification in Thm 3.
1
746
Gene Li retweeted
Data mixing - determining ratios across your training datasets - matters a lot for model quality. While building Olmo 3, we learned it’s hard to set up a method that finds a strong mix, and hard to maintain that mix as datasets change throughout development. Introducing Olmix👇
13
71
9
273
59,487
Thrilled to have contributed to Olmo 3! The best fully open 32B model (data, training recipes, checkpoints and more!) As an intern at AI2 these last 8 months, I’ve grown to deeply appreciate the careful science, iteration, and collaboration that go into models like this and have learned so much from the team. I am more optimistic than ever about the future of open-source and data-centric research right now. My particular contribution was working on the Dolma 3 data mix 👩‍🍳 I was able to apply ideas from some of my earlier mixing work, explore new problem settings, and see firsthand the data challenges that arise when building datasets intended for real models at scale. More on this coming soon!
Announcing Olmo 3, a leading fully open LM suite built for reasoning, chat, & tool use, and an open model flow—not just the final weights, but the entire training journey. Best fully open 32B reasoning model & best 32B base model. 🧵
15
33
9
269
69,955
Check this out! Some fun work with interesting implications for LLM training 🧐
Very satisfied with some neat results on imitation learning. When distribution matching isn’t possible, what’s even the role of demonstrations? Cloning/log-loss minimization? We propose directly encoding reward structure—motivating new algorithmic ideas. arxiv.org/abs/2510.15464
1
9
1,716
Gave a talk at @OpenAI on our work 🌸 POPri “Policy Optimization for Private Data”. POPri is a huge improvement in synthetic data generation under security+privacy constraints! Learn more:
2
2
3
12
2,464