diff --git a/recipes/full_finetune_distributed.py b/recipes/full_finetune_distributed.py
index fc7254b3c4..6a94b0bb38 100644
--- a/recipes/full_finetune_distributed.py
+++ b/recipes/full_finetune_distributed.py
@@ -387,13 +387,15 @@ def setup(self, cfg: DictConfig) -> None:
         )
         if self._compile_optimizer_step:
             if self._optimizer_in_bwd:
-                raise ValueError(
-                    "optimizer_in_bwd not supported with compiling the optimizer step"
+                self._logger.warning(
+                    "Compile optimizer is not supported for optimizer_in_bwd. Setting self._compile_optimizer_step to False."
+                )
+                self._compile_optimizer_step = False
+            else:
+                self._optimizer.step = torch.compile(
+                    self._optimizer.step,
+                    backend=self._compile_backend,
                 )
-            self._optimizer.step = torch.compile(
-                self._optimizer.step,
-                backend=self._compile_backend,
-            )
 
         if self._resume_from_checkpoint:
             # If async checkpointing is enabled, intermediate checkpoints are saved asynchronously
diff --git a/recipes/qat_distributed.py b/recipes/qat_distributed.py
index 1f6b91f163..5432a43adb 100644
--- a/recipes/qat_distributed.py
+++ b/recipes/qat_distributed.py
@@ -387,13 +387,15 @@ def setup(self, cfg: DictConfig) -> None:
         )
         if self._compile_optimizer_step:
             if self._optimizer_in_bwd:
-                raise ValueError(
-                    "optimizer_in_bwd not supported with compiling the optimizer step"
+                self._logger.warning(
+                    "Compile optimizer is not supported for optimizer_in_bwd. Setting self._compile_optimizer_step to False."
+                )
+                self._compile_optimizer_step = False
+            else:
+                self._optimizer.step = torch.compile(
+                    self._optimizer.step,
+                    backend=self._compile_backend,
                 )
-            self._optimizer.step = torch.compile(
-                self._optimizer.step,
-                backend=self._compile_backend,
-            )
 
         if self._resume_from_checkpoint:
             # If async checkpointing is enabled, intermediate checkpoints are saved asynchronously