modeling

modeling_super_linear.py

@@ -1,5 +1,5 @@
 
-from   typing import Optional, Tuple
+from   typing import Optional, Tuple, Union
 import torch, torch.nn as nn, torch.nn.functional as F
 
 from transformers                          import (PreTrainedModel,GenerationMixin,AutoConfig,AutoModelForCausalLM,)
@@ -546,12 +546,11 @@ class superLinear(nn.Module):
     
 "-------------------------------------------------------------------------------------------------------------------"
 class SuperLinearForCausalLM(PreTrainedModel, GenerationMixin):
-    config_class: Tuple[type] = (SuperLinearConfig, SuperLinearConfigFS)
+    config_class: Union[SuperLinearConfig, SuperLinearConfigFS]
 
     def __init__(self, config: Union[SuperLinearConfig, SuperLinearConfigFS]):
         super().__init__(config)
   
-
         # the backbone keeps its own Config dataclass, so build one on‑the‑fly:
         backbone_cfg     = type("Cfg", (), config.to_dict())()
         self.args        = backbone_cfg