MCPcopy Create free account
hub / github.com/Meshcapade/difflocks / forward

Method forward

k_diffusion/models/modules.py:502–508  ·  view source on GitHub ↗
(self, x, cond)

Source from the content-addressed store, hash-verified

500 self.down_proj = apply_wd(zero_init(Linear(d_ff, d_model, bias=False)))
501
502 def forward(self, x, cond):
503 skip = x
504 x = self.norm(x, cond)
505 x = self.up_proj(x)
506 x = self.dropout(x)
507 x = self.down_proj(x)
508 return x + skip
509
510class GlobalTransformerLayer(nn.Module):
511 def __init__(self, d_model, d_ff, d_head, cond_features, dropout=0.0):

Callers

nothing calls this directly

Calls

no outgoing calls

Tested by

no test coverage detected