Randall Balestriero

The Platonic Representation Hypothesis, which holds that separately trained models converge on a shared picture of the world, becomes a theorem for JEPAs (joint-embedding predictive architectures): if each modality's positive pairs link the same world states, the encoders agree up to one rotation. The resulting good-view versus bad-view distinction predicts alignment, and a 109M-parameter text JEPA reaches .184 on the PRH benchmark, above every language model evaluated, up to Qwen2.5-72B at .154.

The Platonic Representation Hypothesis, which holds that separately trained models converge on a shared picture of the world, becomes a theorem for JEPAs (joint-embedding predictive architectures): if each modality's positive pairs link the same world states, the encoders agree up to one rotation. The resulting good-view versus bad-view distinction predicts alignment, and a 109M-parameter text JEPA reaches .184 on the PRH benchmark, above every language model evaluated, up to Qwen2.5-72B at .154.