Make minor improvements to optimizer.py (#1687)

bitsandbytes/optim/optimizer.py

@@ -64,9 +64,9 @@ class GlobalOptimManager:
            parameters (`torch.Tensor` or `list(torch.Tensors)`):
              The input parameters.
            key (`str`):
-             The hyperparamter to override.
+             The hyperparameter to override.
            value:
-             The hyperparameter values.
+             The hyperparameter value.
            key_value_dict (`dict`):
              A dictionary with multiple key-values to override.
 
@@ -115,7 +115,7 @@ class Optimizer8bit(torch.optim.Optimizer):
         Base 8-bit optimizer class.
 
         Arguments:
-            params (`torch.tensor`):
+            params (`torch.Tensor`):
                 The input parameters to optimize.
             optim_bits (`int`, defaults to 32):
                 The number of bits of the optimizer state.
@@ -291,7 +291,7 @@ class Optimizer8bit(torch.optim.Optimizer):
                 self.update_step(group, p, gindex, pindex)
                 torch.cuda.synchronize()
         if self.is_paged:
-            # all paged operation are asynchronous, we need
+            # all paged operations are asynchronous, we need
             # to sync to make sure all tensors are in the right state
             torch.cuda.synchronize()
 
@@ -371,7 +371,7 @@ class Optimizer2State(Optimizer8bit):
         Arguments:
             optimizer_name (`str`):
                 The name of the optimizer.
-            params (`torch.tensor`):
+            params (`torch.Tensor`):
                 The input parameters to optimize.
             lr (`float`, defaults to 1e-3):
                 The learning rate.
@@ -428,7 +428,6 @@ class Optimizer2State(Optimizer8bit):
         if args is None:
             args = {}
             args["optim_bits"] = optim_bits
-            args["percentile_clipping"] = 100
             args["min_8bit_size"] = min_8bit_size
             args["percentile_clipping"] = percentile_clipping
             args["block_wise"] = block_wise
@@ -613,7 +612,7 @@ class Optimizer1State(Optimizer8bit):
         Arguments:
             optimizer_name (`str`):
                 The name of the optimizer.
-            params (`torch.tensor`):
+            params (`torch.Tensor`):
                 The input parameters to optimize.
             lr (`float`, defaults to 1e-3):
                 The learning rate.
@@ -655,7 +654,6 @@ class Optimizer1State(Optimizer8bit):
         if args is None:
             args = {}
             args["optim_bits"] = optim_bits
-            args["percentile_clipping"] = 100
             args["min_8bit_size"] = min_8bit_size
             args["percentile_clipping"] = percentile_clipping
             args["block_wise"] = block_wise