diff --git a/examples/lm_finetuning/finetune_on_pregenerated.py b/examples/lm_finetuning/finetune_on_pregenerated.py
index 1638b02a6f..400be6cdd2 100644
--- a/examples/lm_finetuning/finetune_on_pregenerated.py
+++ b/examples/lm_finetuning/finetune_on_pregenerated.py
@@ -315,7 +315,7 @@ def main():
                     if args.fp16:
                         # modify learning rate with special warm up BERT uses
                         # if args.fp16 is False, BertAdam is used that handles this automatically
-                        lr_this_step = args.learning_rate * warmup_linear.get_lr(global_step/num_train_optimization_steps,
+                        lr_this_step = args.learning_rate * warmup_linear.get_lr(global_step,
                                                                                  args.warmup_proportion)
                         for param_group in optimizer.param_groups:
                             param_group['lr'] = lr_this_step
diff --git a/examples/lm_finetuning/simple_lm_finetuning.py b/examples/lm_finetuning/simple_lm_finetuning.py
index 6511ead590..da01daf4f0 100644
--- a/examples/lm_finetuning/simple_lm_finetuning.py
+++ b/examples/lm_finetuning/simple_lm_finetuning.py
@@ -603,7 +603,7 @@ def main():
                     if args.fp16:
                         # modify learning rate with special warm up BERT uses
                         # if args.fp16 is False, BertAdam is used that handles this automatically
-                        lr_this_step = args.learning_rate * warmup_linear.get_lr(global_step/num_train_optimization_steps,
+                        lr_this_step = args.learning_rate * warmup_linear.get_lr(global_step,
                                                                                  args.warmup_proportion)
                         for param_group in optimizer.param_groups:
                             param_group['lr'] = lr_this_step
diff --git a/examples/run_classifier.py b/examples/run_classifier.py
index 89ab96f50a..17d8f72c37 100644
--- a/examples/run_classifier.py
+++ b/examples/run_classifier.py
@@ -854,7 +854,7 @@ def main():
                     if args.fp16:
                         # modify learning rate with special warm up BERT uses
                         # if args.fp16 is False, BertAdam is used that handles this automatically
-                        lr_this_step = args.learning_rate * warmup_linear.get_lr(global_step/num_train_optimization_steps,
+                        lr_this_step = args.learning_rate * warmup_linear.get_lr(global_step,
                                                                                  args.warmup_proportion)
                         for param_group in optimizer.param_groups:
                             param_group['lr'] = lr_this_step
diff --git a/examples/run_squad.py b/examples/run_squad.py
index c3fdf03774..c7b93ac1dd 100644
--- a/examples/run_squad.py
+++ b/examples/run_squad.py
@@ -1015,7 +1015,7 @@ def main():
                     if args.fp16:
                         # modify learning rate with special warm up BERT uses
                         # if args.fp16 is False, BertAdam is used and handles this automatically
-                        lr_this_step = args.learning_rate * warmup_linear.get_lr(global_step/num_train_optimization_steps,
+                        lr_this_step = args.learning_rate * warmup_linear.get_lr(global_step,
                                                                                  args.warmup_proportion)
                         for param_group in optimizer.param_groups:
                             param_group['lr'] = lr_this_step
diff --git a/examples/run_swag.py b/examples/run_swag.py
index 89f4bdf868..eb5c869553 100644
--- a/examples/run_swag.py
+++ b/examples/run_swag.py
@@ -466,7 +466,7 @@ def main():
                     if args.fp16:
                         # modify learning rate with special warm up BERT uses
                         # if args.fp16 is False, BertAdam is used that handles this automatically
-                        lr_this_step = args.learning_rate * warmup_linear.get_lr(global_step/num_train_optimization_steps,
+                        lr_this_step = args.learning_rate * warmup_linear.get_lr(global_step,
                                                                                  args.warmup_proportion)
                         for param_group in optimizer.param_groups:
                             param_group['lr'] = lr_this_step