fix accelerator prepare during eval only mode (huggingface#24014)

* fix mixed precision prep during eval only mode

* update to address comments

* update to reflect the changes in accelerate

src/transformers/trainer.py

@@ -3141,14 +3141,30 @@ def evaluation_loop(
 
         prediction_loss_only = prediction_loss_only if prediction_loss_only is not None else args.prediction_loss_only
 
-        # if eval is called w/o train init deepspeed here
+        # if eval is called w/o train, handle model prep here
         if self.is_deepspeed_enabled and self.model_wrapped is self.model:
             _, _ = deepspeed_init(self, num_training_steps=0, inference=True)
-            model = self.accelerator.prepare(self.model)
-            self.model_wrapped = self.deepspeed = model
 
         model = self._wrap_model(self.model, training=False, dataloader=dataloader)
 
+        if len(self.accelerator._models) == 0 and model is self.model:
+            model = (
+                self.accelerator.prepare(model)
+                if self.is_deepspeed_enabled
+                else self.accelerator.prepare_model(model, evaluation_mode=True)
+            )
+
+            if self.is_fsdp_enabled:
+                self.model = model
+
+            # for the rest of this function `model` is the outside model, whether it was wrapped or not
+            if model is not self.model:
+                self.model_wrapped = model
+
+            # backward compatibility
+            if self.is_deepspeed_enabled:
+                self.deepspeed = self.model_wrapped
+
         # if full fp16 or bf16 eval is wanted and this ``evaluation`` or ``predict`` isn't called
         # while ``train`` is running, cast it to the right dtype first and then put on device
         if not self.is_in_train:
@@ -3736,14 +3752,30 @@ def prediction_loop(
 
         prediction_loss_only = prediction_loss_only if prediction_loss_only is not None else args.prediction_loss_only
 
-        # if eval is called w/o train init deepspeed here
+        # if eval is called w/o train, handle model prep here
         if self.is_deepspeed_enabled and self.model_wrapped is self.model:
             _, _ = deepspeed_init(self, num_training_steps=0, inference=True)
-            model = self.accelerator.prepare(self.model)
-            self.model_wrapped = self.deepspeed = model
 
         model = self._wrap_model(self.model, training=False, dataloader=dataloader)
 
+        if len(self.accelerator._models) == 0 and model is self.model:
+            model = (
+                self.accelerator.prepare(model)
+                if self.is_deepspeed_enabled
+                else self.accelerator.prepare_model(model, evaluation_mode=True)
+            )
+
+            if self.is_fsdp_enabled:
+                self.model = model
+
+            # for the rest of this function `model` is the outside model, whether it was wrapped or not
+            if model is not self.model:
+                self.model_wrapped = model
+
+            # backward compatibility
+            if self.is_deepspeed_enabled:
+                self.deepspeed = self.model_wrapped
+
         # if full fp16 or bf16 eval is wanted and this ``evaluation`` or ``predict`` isn't called
         # while ``train`` is running, cast it to the right dtype first and then put on device
         if not self.is_in_train: