Update app.py

app.py

@@ -329,7 +329,7 @@ pipeline = LTX23DistilledA2VPipeline(
     spatial_upsampler_path=spatial_upsampler_path,
     gemma_root=gemma_root,
     loras=[],
-    quantization=None,  # keep FP8 quantization unchanged
+    quantization=QuantizationPolicy.fp8_cast(),
 )
 # ----------------------------------------------------------------
 
@@ -428,7 +428,7 @@ def prepare_lora_cache(
             spatial_upsampler_path=str(spatial_upsampler_path),
             gemma_root_path=str(gemma_root),
             loras=tuple(loras_for_builder),
-            quantization=None,
+            quantization=QuantizationPolicy.fp8_cast(),
         )
         new_transformer_cpu = tmp_ledger.transformer()