[diffusion] fix: fix RowParallel LoRA merged forwarding (#24410)

python/sglang/multimodal_gen/runtime/layers/lora/linear.py

@@ -513,6 +513,9 @@ class RowParallelLinearWithLoRA(BaseLayerWithLoRA):
         super().__init__(base_layer, lora_rank, lora_alpha)
 
     def forward(self, input_: torch.Tensor):
+        if self.merged or self.disable_lora:
+            return self.base_layer(input_)
+
         lora_A = self.lora_A
         lora_B = self.lora_B
         if isinstance(self.lora_B, DTensor):

python/sglang/multimodal_gen/test/test_utils.py

@@ -33,7 +33,7 @@ if TYPE_CHECKING:
 
 logger = init_logger(__name__)
 
-SGL_TEST_FILES_CI_DATA_REVISION = "437539b330592b1421d239b22d7d76b4a6d08fda"
+SGL_TEST_FILES_CI_DATA_REVISION = "4d9eff3b05b0ffe1d3529e8bb148b63af11a4b92"
 SGL_TEST_FILES_CONSISTENCY_GT_ROOT = (
     "https://raw.githubusercontent.com/"
     f"sgl-project/ci-data/{SGL_TEST_FILES_CI_DATA_REVISION}/"