fix resuming from checkpoint (#2001)

examples/by_feature/deepspeed_with_config_support.py

@@ -602,15 +602,22 @@ def group_texts(examples):
             resume_step -= starting_epoch * num_update_steps_per_epoch
             completed_steps = resume_step
 
+    # update progress bar if resumed from checkpoint
+    progress_bar.update(completed_steps)
+
     for epoch in range(starting_epoch, args.num_train_epochs):
         model.train()
         if args.with_tracking:
             total_loss = 0
 
         # skip new `skip_first_batches` to skip the batches when resuming from ckpt
-        if args.resume_from_checkpoint:
-            train_dataloader = accelerator.skip_first_batches(train_dataloader, num_batches=resume_step)
-        for step, batch in enumerate(train_dataloader):
+        if args.resume_from_checkpoint and epoch == starting_epoch and resume_step is not None:
+            # We need to skip steps until we reach the resumed step
+            active_dataloader = accelerator.skip_first_batches(train_dataloader, resume_step)
+        else:
+            # After the first iteration though, we need to go back to the original dataloader
+            active_dataloader = train_dataloader
+        for step, batch in enumerate(active_dataloader):
             # In particular, DeepSpeed handles `gradient_accumulation` via `DeepSpeedEngine`.
             # Below, we use `accelerator.accumulate` if the user
             # wants to switch to other approaches such as plain DDP, PyTorch FSDP ...