modality scale set to 1

prishajain1 · prishajain1 · commit b99a37f1c738 · 2026-04-12T23:39:20.000+05:30
diff --git a/src/maxdiffusion/configs/ltx2_3_video.yml b/src/maxdiffusion/configs/ltx2_3_video.yml
@@ -33,8 +33,8 @@ audio_guidance_scale: 7.0
 audio_guidance_rescale: 0.7
 stg_scale: 1.0
 audio_stg_scale: 1.0
-modality_scale: 3.0
-audio_modality_scale: 3.0
+modality_scale: 1.0
+audio_modality_scale: 1.0
 use_cross_timestep: true
 spatio_temporal_guidance_blocks: [28]
 fps: 24
diff --git a/src/maxdiffusion/pipelines/ltx2/ltx2_pipeline.py b/src/maxdiffusion/pipelines/ltx2/ltx2_pipeline.py
@@ -111,7 +111,6 @@ def create_model(rngs: nnx.Rngs, ltx2_config: dict):
     transformer = LTX2VideoTransformer3DModel(
         **ltx2_config,
         spatio_temporal_guidance_blocks=tuple(getattr(config, "spatio_temporal_guidance_blocks", ())),
-        perturbed_attn=getattr(config, "model_name", "") == "ltx2.3",
         rngs=rngs
     )
     return transformer

Original file line number	Diff line number	Diff line change
`@@ -111,7 +111,6 @@ def create_model(rngs: nnx.Rngs, ltx2_config: dict):`
`111`	`111`	`transformer = LTX2VideoTransformer3DModel(`
`112`	`112`	`**ltx2_config,`
`113`	`113`	`spatio_temporal_guidance_blocks=tuple(getattr(config, "spatio_temporal_guidance_blocks", ())),`
`114`		`- perturbed_attn=getattr(config, "model_name", "") == "ltx2.3",`
`115`	`114`	`rngs=rngs`
`116`	`115`	`)`
`117`	`116`	`return transformer`