Denoiser › 12 denoiser blocks
Hidden latents, next block
×tensor
The residual features after both mechanism updates.
- shape
- [B batch, T positions, 768 d]
Sources
Contextual evidence from Denoiser block
A block owns two gain-only LayerNorms, a fused query–key–value projection and an output projection without bias, an MLP of the configured kind, and, with modulation, an affine layer from the time condition to six vectors, initialized at zero.
The implementation, pinned at revision e18c42acdfb7, not yet public
Contextual evidence from Denoiser block
A block normalizes, modulates by the time condition’s shift and scale, projects to queries, keys and values, rotates queries and keys, attends over all positions with softcap, and adds the projected output scaled by a gate; the MLP branch does the same with its own shift, scale and gate.
The implementation, pinned at revision e18c42acdfb7, not yet public
Contextual evidence from Denoiser block
The denoiser’s stack applies the blocks in order and then the output layer, each with the time condition.
The implementation, pinned at revision e18c42acdfb7, not yet public
Contextual evidence from Denoiser block
Diffusion: self-conditioning on, `t_min` 0.05 and `eps` 10⁻⁵; the adaptive time sampler has 100 bins, a minimum weight of 0.0001, an EMA of 0.9 and updates every 5,000 steps; the denoiser’s EMA decay is 0.9999; scheduler `tanh` with d = 3, solver `euler`, `ode_sampling` false, N = 128 by default; the denoiser has 12 blocks of width 768 with 12 heads, time conditioning `adaln`, a standard MLP, softcap 50 and self-conditioning by concatenation; the timestep embedding’s `max_period` is 10.
The implementation, pinned at revision e18c42acdfb7, not yet public
Contextual evidence from Add MLP update
The residual update is scale · dropout(x) added to the residual; modulation is x · (1 + scale) + shift.
The implementation, pinned at revision e18c42acdfb7, not yet public