Merge branch 'save_checkpoint_fix' into 'main'

Add missing num_floating_point_operations_so_far argument to save_checkpoint_and_time call

See merge request ADLR/megatron-lm!1115

(cherry picked from commit 0f0279a)

2fb398c Add missing num_floating_point_operations_so_far argument to save_checkpoint_and_time call

Author: jaredcasper

megatron/training.py

@@ -954,7 +954,8 @@ def track_e2e_metrics():
             assert get_num_microbatches() > num_microbatches, \
                 "number of microbatches should be increasing due to batch size rampup"
             save_checkpoint_and_time(iteration, model, optimizer,
-                                     opt_param_scheduler)
+                                     opt_param_scheduler,
+                                     num_floating_point_operations_so_far)
         num_microbatches = get_num_microbatches()
         update_num_microbatches(args.consumed_train_samples, consistency_check=True)