MCPcopy Index your code
hub / github.com/MotrixLab/MotionDiffuse / __init__

Method __init__

text2motion/models/transformer.py:231–240  ·  view source on GitHub ↗
(self, seq_len, latent_dim, text_latent_dim, num_head, dropout, time_embed_dim)

Source from the content-addressed store, hash-verified

229class TemporalCrossAttention(nn.Module):
230
231 def __init__(self, seq_len, latent_dim, text_latent_dim, num_head, dropout, time_embed_dim):
232 super().__init__()
233 self.num_head = num_head
234 self.norm = nn.LayerNorm(latent_dim)
235 self.text_norm = nn.LayerNorm(text_latent_dim)
236 self.query = nn.Linear(latent_dim, latent_dim)
237 self.key = nn.Linear(text_latent_dim, latent_dim)
238 self.value = nn.Linear(text_latent_dim, latent_dim)
239 self.dropout = nn.Dropout(dropout)
240 self.proj_out = StylizationBlock(latent_dim, time_embed_dim, dropout)
241
242 def forward(self, x, xf, emb):
243 """

Callers

nothing calls this directly

Calls 2

StylizationBlockClass · 0.85
__init__Method · 0.45

Tested by

no test coverage detected