Our Mission
A poem is a shadow of the act of writing poetry.
Humanity casts many shadows. All literature, letters, recipes, and tweets. Mathematical proofs and git repositories. Laws, treaties, and declarations of war. Financial statements, employee performance reports, and bankruptcy filings. Podcasts, operas, accidental voicemails, YouTube videos and Hollywood blockbusters. Restaurant reviews and love letters and times tables. Individually, each of these pieces of information is nothing but the faintest shadow of the process that produced it. But collectively, these shadows tell a rich story about the world.
Since the dawn of humanity, the brain alone could reconstruct the world from these shadows. But the last decade of deep learning has convinced us that this won’t be the case for much longer. Though significant challenges remain, we stand poised to solve them. If successful, it will become possible to synthesize every documentable aspect of humanity into the weights of a neural network.
Our mission is to train a neural network to model all human output.
There are two primary challenges in our pursuit of this goal.
- Currently, it’s not technically feasible to train a model that can ingest all the data that we can collect. Limitations around context length, modality, and throughput force us to use only a small subset of the data available to us.
- Much of the data we will need has never been collected, curated, and organized into datasets that we can use for training.
We are building a world-class team of engineers and researchers to tackle these challenges, united around shared principles and a specific research agenda. We value clear thinking, sharing knowledge, and an extreme commitment to scientific honesty. Our research is guided by mathematical beauty and grounded in rigorous empiricism. We are committed to letting the quality of our work speak for itself. No hype, no fluff. All meat.
If this vision resonates, please reach out: 777b7a607577605479757a7d72716760757d3a777b79