MCPcopy Create free account
hub / github.com/tdrussell/diffusion-pipe / forward

Method forward

models/sdxl.py:809–814  ·  view source on GitHub ↗
(self, inputs)

Source from the content-addressed store, hash-verified

807
808 @torch.autocast('cuda', dtype=AUTOCAST_DTYPE)
809 def forward(self, inputs):
810 hidden_states, timesteps, emb, encoder_hidden_states, *res_hidden_states, forward_upsample_size = inputs
811 hidden_states = self.resnet(hidden_states, emb)
812 if self.attn is not None:
813 hidden_states = self.attn(hidden_states, encoder_hidden_states=encoder_hidden_states, return_dict=False)[0]
814 return make_contiguous(hidden_states, timesteps, emb, encoder_hidden_states, *res_hidden_states, forward_upsample_size)
815
816
817class UpBlockInnerLayer(nn.Module):

Callers

nothing calls this directly

Calls 1

make_contiguousFunction · 0.90

Tested by

no test coverage detected