diff --git a/src/transformers/trainer.py b/src/transformers/trainer.py
index 1f7df7e9f3b0..bf0538516864 100755
--- a/src/transformers/trainer.py
+++ b/src/transformers/trainer.py
@@ -1616,7 +1616,7 @@ def train(
             if resume_from_checkpoint is None:
                 raise ValueError(f"No valid checkpoint found in output directory ({args.output_dir})")
 
-        if resume_from_checkpoint is not None and not is_sagemaker_mp_enabled():
+        if resume_from_checkpoint is not None and not is_sagemaker_mp_enabled() and args.deepspeed is None:
             self._load_from_checkpoint(resume_from_checkpoint)
 
         # If model was re-initialized, put it on the right device and update self.model_wrapped