hf_model.py: small refactoring (#242)

* hf_model.py: encoder-decoder fix partial locking

* refactor ifs, less ifs

* try roms method

* fix

* back to this

* update

* revert

src/open_clip/hf_model.py

@@ -85,19 +85,13 @@ def __init__(
             raise RuntimeError("Please `pip install transformers` to use pre-trained HuggingFace models")
         if config is None:
             self.config = AutoConfig.from_pretrained(model_name_or_path)
-            if pretrained:
-                # TODO: do all model configs have this attribute? PretrainedConfig does so yes??
-                if hasattr(self.config, "is_encoder_decoder") and self.config.is_encoder_decoder:
-                    self.transformer = AutoModel.from_pretrained(model_name_or_path)
-                    self.transformer = self.transformer.encoder
-                else:
-                    self.transformer = AutoModel.from_pretrained(model_name_or_path, add_pooling_layer=uses_transformer_pooler)
+            create_func, model_args = (AutoModel.from_pretrained, model_name_or_path) if pretrained else (AutoModel.from_config, self.config)
+            # TODO: do all model configs have this attribute? PretrainedConfig does so yes??
+            if hasattr(self.config, "is_encoder_decoder") and self.config.is_encoder_decoder:
+                self.transformer = create_func(model_args)
+                self.transformer = self.transformer.encoder
             else:
-                if hasattr(self.config, "is_encoder_decoder") and self.config.is_encoder_decoder:
-                    self.transformer = AutoModel.from_config(self.config)
-                    self.transformer = self.transformer.encoder
-                else:
-                    self.transformer = AutoModel.from_config(self.config, add_pooling_layer=uses_transformer_pooler)
+                self.transformer = create_func(model_args, add_pooling_layer=uses_transformer_pooler)
         else:
             self.config = config
             self.transformer = AutoModel.from_config(config)