fix loading HF model

.pre-commit-config.yaml

@@ -36,7 +36,7 @@ repos:
       - id: detect-private-key
 
   - repo: https://github.com/asottile/pyupgrade
-    rev: v2.38.2
+    rev: v3.2.2
     hooks:
       - id: pyupgrade
         args: [--py37-plus]
@@ -56,7 +56,7 @@ repos:
         require_serial: false
 
   - repo: https://github.com/psf/black
-    rev: 22.8.0
+    rev: 22.10.0
     hooks:
       - id: black
         name: Format code

CHANGELOG.md

@@ -9,6 +9,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- Fixed loading HF model ([#306](https://github.com/Lightning-AI/lightning-transformers/pull/306))
+
+
 - Fixed passing config name to `CNNDailyMailSummarizationDataModule` ([#310](https://github.com/Lightning-AI/lightning-transformers/pull/310))
 
 

lightning_transformers/core/model.py

@@ -76,7 +76,9 @@ def initialize_model(self, pretrained_model_name_or_path: str):
                 pretrained_model_name_or_path, **self.model_data_kwargs
             )
         else:
-            config = AutoConfig.from_pretrained(pretrained_model_name_or_path, **self.model_data_kwargs)
+            config = AutoConfig.from_pretrained(
+                pretrained_model_name_or_path=self.model_data_kwargs["pretrained_model"], **self.model_data_kwargs
+            )
             self.model = self.downstream_model_type.from_config(config)
 
     def configure_optimizers(self) -> Dict: