Research / Slide 02
Not another attention layer.
A field-based route.
Tokens write into a shared medium, waves propagate through the field, heads couple during the process, and token positions read the result back out.
ScatterTokens deposit value information onto a one-dimensional continuous field using bilinear interpolation.
Research / Slide 03
Each head becomes a damped wave.
Instead of forming an attention matrix, Wave Field uses a causal oscillator kernel with three learnable scalars per head.
Frequency controls pattern scale, damping controls attention range, and phase gives heads different responses. The same compact equation can express local syntax and longer document structure.
k(t) = exp(-0.45t) cos(2.00t + 0.00), t >= 0