Position: Weight Space Should Be a First-Class Generative AI Modality

In three linesNeural network weights constitute a first-class data modality. This position paper proposes treating checkpoints as generative data: synthesizing weights on demand matches or exceeds fine-tuning while reducing adaptation costs by orders of magnitude. High-performing models occupy structured regions of weight space (symmetry, modularity, shared subspaces).

Summary generated by Claude — human-verified