Rigorous AI training data
At π1 Labs, we create specialized training data to detect critical holes in the reasoning of AI models. Our team of passionate mathematicians and physicists from leading institutions develops original problems with difficulties ranging from that of high school competition problems to advanced graduate and research. Each problem is designed to target and expose logical weaknesses, is rigorously reviewed, and includes a detailed solution. We believe that the future of AI in the basic sciences is tied to the systematic analysis and understanding of the strengths and weaknesses of models combined with careful human oversight. We strive to be an important part of a serious integration of AI into research, and see great potential in increasing the precision of AI. We intend to expand into other basic sciences while maintaining the same high quality.
Datasets
Our datasets cover a wide range of topics in mathematics and physics. To enhance AI training, some problems are multi-modal, incorporating text with diagrams and images. Our carefully crafted problems allow AI systems to engage with questions at all levels.
We produce problems at the high school, undergraduate, graduate, and research levels, with a specialization in advanced topics. Our teams have members with expert knowledge in a wide range of advanced fields in mathematics and physics, from analysis to algebraic geometry and quantum field theory to cosmology, covering all core areas in addition to niche areas.
Request Samples
For samples, please fill out the form below. You could also use the contact form further down to request a meeting to discuss how we could meet your dataset needs.
About Us
π1 Labs started as a group of mathematicians from top universities aiming to push the boundaries of AI capabilities, and later expanded to include physicists. Our team consists of active researchers as well as medalists from and trainers for the International Mathematical Olympiad (IMO) and International Physics Olympiad (IPhO). We are committed to progress in the basic sciences in the AI age. If our work interests you, we invite you to get in touch.