fix: loading from pretrained

modeling_hf_nomic_bert.py

@@ -330,7 +330,10 @@ class NomicBertPreTrainedModel(PreTrainedModel):
         if "add_pooling_layer" in kwargs:
             model = cls(config, *inputs, add_pooling_layer=kwargs.pop("add_pooling_layer"))
         else:
-            model = cls(config, *inputs, add_pooling_layer=False)
+            if cls == NomicBertModel:
+                model = cls(config, *inputs, add_pooling_layer=False)
+            else:
+                model = cls(config, *inputs)
         # TODO: fix this
         # Assuming we know what we're doing when loading from disk
         # Prob a bad assumption but i'm tired and want to train this asap