[deepspeed] replaced deprecated init arg (#13587)

* [deepspeed] replaced deprecated init arg

* Trigger CI

setup.py

@@ -90,7 +90,7 @@ _deps = [
     "cookiecutter==1.7.2",
     "dataclasses",
     "datasets",
-    "deepspeed>=0.5.1",
+    "deepspeed>=0.5.3",
     "docutils==0.16.0",
     "fairscale>0.3",
     "faiss-cpu",

src/transformers/dependency_versions_table.py

@@ -8,7 +8,7 @@ deps = {
     "cookiecutter": "cookiecutter==1.7.2",
     "dataclasses": "dataclasses",
     "datasets": "datasets",
-    "deepspeed": "deepspeed>=0.5.1",
+    "deepspeed": "deepspeed>=0.5.3",
     "docutils": "docutils==0.16.0",
     "fairscale": "fairscale>0.3",
     "faiss-cpu": "faiss-cpu",

src/transformers/modeling_utils.py

@@ -492,7 +492,7 @@ class PreTrainedModel(nn.Module, ModuleUtilsMixin, GenerationMixin, PushToHubMix
             logger.info("Detected DeepSpeed ZeRO-3: activating zero.init() for this model")
             # this immediately partitions the model across all gpus, to avoid the overhead in time
             # and memory copying it on CPU or each GPU first
-            with deepspeed.zero.Init(config=deepspeed_config()):
+            with deepspeed.zero.Init(config_dict_or_path=deepspeed_config()):
                 model = cls(config, **kwargs)
         else:
             model = cls(config, **kwargs)