Bring back PartialState DeepSpeed (#22921)

* Bring back deepspeed integration

* Branchname

* Self-scheduled

* newline

* Use deepspeed env var

* Remove comment

* Del env var after partialstate

src/transformers/training_args.py

@@ -1544,39 +1544,26 @@ class TrainingArguments:
             self._n_gpu = 1
             torch.cuda.set_device(device)
         elif self.deepspeed:
-            # deepspeed inits torch.distributed internally
-            from .deepspeed import is_deepspeed_available
-
-            if not is_deepspeed_available():
-                raise ImportError("--deepspeed requires deepspeed: `pip install deepspeed`.")
-            import deepspeed
-
-            deepspeed.init_distributed(timeout=timedelta(seconds=self.ddp_timeout))
-
-            # workaround for setups like notebooks where the launcher can't be used,
-            # but deepspeed requires a dist env.
-            # env LOCAL_RANK could be set manually by the user, or via init_distributed if mpi4py is installed
-            self.local_rank = int(os.environ.get("LOCAL_RANK", "-1"))
-
-            device = torch.device("cuda", self.local_rank)
+            # Need to do similar for Accelerator init
+            os.environ["ACCELERATE_USE_DEEPSPEED"] = "true"
+            self.distributed_state = PartialState(timeout=timedelta(seconds=self.ddp_timeout))
+            del os.environ["ACCELERATE_USE_DEEPSPEED"]
             self._n_gpu = 1
         else:
             self.distributed_state = PartialState(backend=self.xpu_backend)
             self._n_gpu = 1
-        if not is_sagemaker_mp_enabled() and not self.deepspeed:
+        if not is_sagemaker_mp_enabled():
             device = self.distributed_state.device
             self.local_rank = self.distributed_state.local_process_index
         if (
             torch.distributed.is_available()
             and torch.distributed.is_initialized()
-            and hasattr(self, "distributed_state")
             and self.distributed_state.distributed_type == DistributedType.NO
         ):
             logger.warning(
                 "torch.distributed process group is initialized, but parallel_mode != ParallelMode.DISTRIBUTED. "
                 "In order to use Torch DDP, launch your script with `python -m torch.distributed.launch"
             )
-        if not self.deepspeed:
         if is_torch_tpu_available():
             device = self.distributed_state.device
             self._n_gpu = 0
@@ -1615,7 +1602,6 @@ class TrainingArguments:
                 # trigger an error that a device index is missing. Index 0 takes into account the
                 # GPUs available in the environment, so `CUDA_VISIBLE_DEVICES=1,2` with `cuda:0`
                 # will use the first GPU in that env, i.e. GPU#1
-                    # device = self.distributed_state.device
                 device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
                 # Sometimes the line in the postinit has not been run before we end up here, so just checking we're not at
                 # the default value.
@@ -1664,7 +1650,7 @@ class TrainingArguments:
             return ParallelMode.SAGEMAKER_MODEL_PARALLEL
         elif is_sagemaker_dp_enabled():
             return ParallelMode.SAGEMAKER_DATA_PARALLEL
-        elif self.deepspeed or self.distributed_state.distributed_type != DistributedType.NO:
+        elif hasattr(self, "distributed_state") and self.distributed_state.distributed_type != DistributedType.NO:
             return ParallelMode.DISTRIBUTED
         elif self.n_gpu > 1:
             return ParallelMode.NOT_DISTRIBUTED