[CI][Bugfix] Fix LTX audio-video warmup output typing (#3964)

Signed-off-by: david6666666 <530634352@qq.com>

vllm_omni/diffusion/models/ltx2/pipeline_ltx2.py

@@ -158,7 +158,7 @@ class _VideoAudioScheduler:
 
 class LTX2Pipeline(nn.Module, CFGParallelMixin, ProgressBarMixin):
     # Audio is diffused jointly with video; warmup must size audio tokens.
-    support_audio_output = True
+    dummy_run_num_frames = 2
 
     def __init__(
         self,
@@ -1121,6 +1121,8 @@ class LTX2Pipeline(nn.Module, CFGParallelMixin, ProgressBarMixin):
 class LTX2TwoStagesPipeline(nn.Module, SupportsComponentDiscovery):
     """LTX2TwoStagesPipeline is for two stages image to video generation"""
 
+    dummy_run_num_frames = 2
+
     _dit_modules: ClassVar[list[str]] = ["pipe.transformer"]
     _encoder_modules: ClassVar[list[str]] = ["pipe.text_encoder"]
     _vae_modules: ClassVar[list[str]] = ["pipe.vae", "pipe.audio_vae"]

vllm_omni/diffusion/models/ltx2/pipeline_ltx2_3.py

@@ -123,7 +123,7 @@ class LTX23Pipeline(nn.Module, CFGParallelMixin, ProgressBarMixin):
     """
 
     # Audio is diffused jointly with video; warmup must size audio tokens.
-    support_audio_output = True
+    dummy_run_num_frames = 2
 
     def __init__(
         self,

vllm_omni/diffusion/models/ltx2/pipeline_ltx2_image2video.py

@@ -735,6 +735,7 @@ class LTX2ImageToVideoTwoStagesPipeline(nn.Module, SupportsComponentDiscovery):
     """LTXImageToVideoTwoStagesPipeline is for two stages image to video generation"""
 
     support_image_input = True
+    dummy_run_num_frames = 2
 
     _dit_modules: ClassVar[list[str]] = ["pipe.transformer"]
     _encoder_modules: ClassVar[list[str]] = ["pipe.text_encoder"]