cross attn dim and vocoder weights

prishajain1 · prishajain1 · commit 632d2c7565a2 · 2026-04-10T14:11:48.000+05:30
diff --git a/src/maxdiffusion/models/ltx2/transformer_ltx2.py b/src/maxdiffusion/models/ltx2/transformer_ltx2.py
@@ -1166,7 +1166,6 @@ def scan_fn(carry, block_and_mask):
             ca_audio_rotary_emb=audio_cross_attn_rotary_emb,
             encoder_attention_mask=encoder_attention_mask,
             audio_encoder_attention_mask=audio_encoder_attention_mask,
-            perturbation_mask=mask,
         )
       return (
           hidden_states_out.astype(hidden_states.dtype),

Original file line number	Diff line number	Diff line change
`@@ -1166,7 +1166,6 @@ def scan_fn(carry, block_and_mask):`
`1166`	`1166`	`ca_audio_rotary_emb=audio_cross_attn_rotary_emb,`
`1167`	`1167`	`encoder_attention_mask=encoder_attention_mask,`
`1168`	`1168`	`audio_encoder_attention_mask=audio_encoder_attention_mask,`
`1169`		`- perturbation_mask=mask,`
`1170`	`1169`	`)`
`1171`	`1170`	`return (`
`1172`	`1171`	`hidden_states_out.astype(hidden_states.dtype),`