fix

Author: prishajain1

src/maxdiffusion/models/ltx2/transformer_ltx2.py

@@ -699,15 +699,15 @@ def __init__(
           rngs=rngs,
           in_features=self.caption_channels,
           hidden_size=self.cross_attention_dim,
-          embedding_dim=inner_dim,
+          embedding_dim=self.cross_attention_dim,
           dtype=self.dtype,
           weights_dtype=self.weights_dtype,
       )
       self.audio_caption_projection = NNXCombinedTimestepTextProjEmbeddings(
           rngs=rngs,
           in_features=self.audio_caption_channels,
           hidden_size=self.audio_cross_attention_dim,
-          embedding_dim=audio_inner_dim,
+          embedding_dim=self.audio_cross_attention_dim,
           dtype=self.dtype,
           weights_dtype=self.weights_dtype,
       )