Fix pad across processes dim in trainer and not being able to set the timeout (#24775)

* dim, and rm copy

* Don't rm copy for now

* Oops

* pad index

* Should be a working test

* Tickle down ddp timeout

* Put fix back in now that testing locally is done

* Better comment specifying timeout

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

---------

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

src/transformers/trainer.py

@@ -3131,9 +3131,9 @@ class Trainer:
                 losses = self.accelerator.gather_for_metrics((loss.repeat(batch_size)))
                 losses_host = losses if losses_host is None else nested_concat(losses_host, losses, padding_index=-100)
             if labels is not None:
-                labels = self.accelerator.pad_across_processes(labels)
+                labels = self.accelerator.pad_across_processes(labels, dim=1, pad_index=-100)
             if inputs_decode is not None:
-                inputs_decode = self.accelerator.pad_across_processes(inputs_decode)
+                inputs_decode = self.accelerator.pad_across_processes(inputs_decode, dim=1, pad_index=-100)
                 inputs_decode = self.accelerator.gather_for_metrics((inputs_decode))
                 inputs_host = (
                     inputs_decode
@@ -3141,7 +3141,7 @@ class Trainer:
                     else nested_concat(inputs_host, inputs_decode, padding_index=-100)
                 )
             if logits is not None:
-                logits = self.accelerator.pad_across_processes(logits)
+                logits = self.accelerator.pad_across_processes(logits, dim=1, pad_index=-100)
                 if self.preprocess_logits_for_metrics is not None:
                     logits = self.preprocess_logits_for_metrics(logits, labels)
                 logits = self.accelerator.gather_for_metrics((logits))

src/transformers/training_args.py

@@ -1714,7 +1714,9 @@ class TrainingArguments:
             del os.environ["ACCELERATE_USE_DEEPSPEED"]
             self._n_gpu = 1
         else:
-            self.distributed_state = PartialState(backend=self.ddp_backend)
+            self.distributed_state = PartialState(
+                backend=self.ddp_backend, timeout=timedelta(seconds=self.ddp_timeout)
+            )
             self._n_gpu = 1
         if not is_sagemaker_mp_enabled():
             device = self.distributed_state.device

tests/trainer/test_trainer.py

@@ -49,6 +49,7 @@ from transformers.testing_utils import (
     USER,
     CaptureLogger,
     TestCasePlus,
+    execute_subprocess_async,
     get_gpu_count,
     get_tests_dir,
     is_staging_test,
@@ -2098,6 +2099,51 @@ class TrainerIntegrationTest(TestCasePlus, TrainerIntegrationCommon):
         self.assertListEqual(trainer.optimizer.param_groups[0]["params"], wd_params)
         self.assertListEqual(trainer.optimizer.param_groups[1]["params"], no_wd_params)
 
+    @slow
+    @require_torch_multi_gpu
+    def test_end_to_end_example(self):
+        # Tests that `translation.py` will run without issues
+        script_path = os.path.abspath(
+            os.path.join(
+                os.path.dirname(__file__), "..", "..", "examples", "pytorch", "translation", "run_translation.py"
+            )
+        )
+
+        with tempfile.TemporaryDirectory() as tmpdir:
+            command = [
+                "accelerate",
+                "launch",
+                script_path,
+                "--model_name_or_path",
+                "t5-small",
+                "--per_device_train_batch_size",
+                "1",
+                "--output_dir",
+                tmpdir,
+                "--overwrite_output_dir",
+                "--do_train",
+                "--max_train_samples",
+                "64",
+                "--num_train_epochs",
+                "1",
+                "--dataset_name",
+                "wmt16",
+                "--dataset_config",
+                "ro-en",
+                "--source_lang",
+                "en",
+                "--target_lang",
+                "ro",
+                "--do_predict",
+                "--max_predict_samples",
+                "64",
+                "--predict_with_generate",
+                "--ddp_timeout",
+                "60",
+            ]
+            execute_subprocess_async(command)
+            # successful return here == success - any errors would have caused an error or a timeout in the sub-call
+
 
 @require_torch
 @is_staging_test