diff --git a/deepspeed/runtime/engine.py b/deepspeed/runtime/engine.py
index dd77413c82a2..6d8040fe8ec2 100755
--- a/deepspeed/runtime/engine.py
+++ b/deepspeed/runtime/engine.py
@@ -1662,27 +1662,19 @@ def _save_checkpoint(self, save_dir, tag, client_state={}):
         # then instead just returns None.
         self._curr_ckpt_path = os.path.join(save_dir, tag)
 
-        state = {
-            'module':
-            self.module_state_dict(),
-            'optimizer':
-            self.optimizer.state_dict()
+        state = dict(
+            module=self.module_state_dict(),
+            optimizer=self.optimizer.state_dict()
             if self.optimizer and not self.zero_optimization() else None,
-            'lr_scheduler':
-            self.lr_scheduler.state_dict() if self.lr_scheduler is not None else None,
-            'csr_tensor_module_names':
-            self.csr_tensor_module_names,
-            'skipped_steps':
-            self.skipped_steps,
-            'global_steps':
-            self.global_steps,
-            'global_samples':
-            self.global_samples,
-            'dp_world_size':
-            self.dp_world_size,
-            'mp_world_size':
-            self.mp_world_size
-        }
+            lr_scheduler=self.lr_scheduler.state_dict()
+            if self.lr_scheduler is not None else None,
+            csr_tensor_module_names=self.csr_tensor_module_names,
+            skipped_steps=self.skipped_steps,
+            global_steps=self.global_steps,
+            global_samples=self.global_samples,
+            dp_world_size=self.dp_world_size,
+            mp_world_size=self.mp_world_size,
+        )
         state.update(client_state)
 
         log_dist(message=f'Saving model checkpoint: {save_path}', ranks=[0])
diff --git a/deepspeed/runtime/zero/config.py b/deepspeed/runtime/zero/config.py
index 622ffa9ba1cb..c179d01f2988 100755
--- a/deepspeed/runtime/zero/config.py
+++ b/deepspeed/runtime/zero/config.py
@@ -36,12 +36,6 @@ def __init__(self, param_dict):
         self.max_reuse_distance = None
         self.gather_fp16_weights_on_model_save = None
 
-        #Stage3 Specific Parameters
-        self.prefetch_bucket_size = None
-        self.param_persistence_threshold = None
-        self.max_live_parameters = None
-        self.max_reuse_distance = None
-
         if ZERO_OPTIMIZATION in param_dict.keys():
             zero_config_dict = param_dict[ZERO_OPTIMIZATION]
             if type(zero_config_dict) is bool:
diff --git a/deepspeed/runtime/zero/partition_parameters.py b/deepspeed/runtime/zero/partition_parameters.py
index c654e66abc59..0acc675985ca 100755
--- a/deepspeed/runtime/zero/partition_parameters.py
+++ b/deepspeed/runtime/zero/partition_parameters.py
@@ -399,10 +399,10 @@ def _convert_to_deepspeed_param(self, param):
         # Stores the shape of the original tensor
         param.ds_shape = param.shape
 
-        # Stores the number of elements in the original parmaeter without padding
+        # Stores the number of elements in the original parameter without padding
         param.ds_numel = param.numel()
 
-        # Stores the paritioned copy of the tensor
+        # Stores the partitioned copy of the tensor
         param.ds_tensor = None
 
         # Keeps track of how many active sub-modules need this param at any given point in time
diff --git a/deepspeed/runtime/zero/stage3.py b/deepspeed/runtime/zero/stage3.py
index ea4653578616..e5299949fcf6 100755
--- a/deepspeed/runtime/zero/stage3.py
+++ b/deepspeed/runtime/zero/stage3.py
@@ -580,7 +580,7 @@ def __init__(self,
                  gradient_accumulation_steps=1,
                  elastic_checkpoint=False):
 
-        see_memory_usage("Stage 3 intialize beginning", force=True)
+        see_memory_usage("Stage 3 initialize beginning", force=True)
 
         if dist.get_rank() == 0:
             logger.info(f"Reduce bucket size {reduce_bucket_size}")