fix(modeling): use correct `base_model_prefix` name

modeling_stablelm_epoch.py

@@ -542,7 +542,7 @@ class StableLMEpochPreTrainedModel(PreTrainedModel):
     """
 
     config_class = StableLMEpochConfig
-    base_model_prefix = "transformer"
+    base_model_prefix = "model"
     supports_gradient_checkpointing = True
     _no_split_modules = ["DecoderLayer"]
     _skip_keys_device_placement = "past_key_values"