fixes

configuration_ministu.py

@@ -4,31 +4,31 @@ from transformers import PretrainedConfig, AutoConfig
 class MiniSTUConfig(PretrainedConfig):
     model_type = "ministu"
 
-        def __init__(
-        self,
-        bsz: int = 1,
-        dim: int = 896,
-        num_heads: int = 8,
-        num_layers: int = 12,
-        seq_len: int = 8192,
-        weight_tying: bool = False,
-        window_size: int = 1024,
-        vocab_size: int = 200064,
-        mlp_scale: int = 12,
-        bias: bool = False,
-        dropout: float = 0.0,
-        num_eigh: int = 24,
-        use_hankel_L: bool = False,
-        use_flash_fft: bool = True,
-        use_approx: bool = True,
-        use_attn: bool = True,
-        softcap: float = 50.0,
-        theta: float = 10_000.0,
-        use_alibi: bool = False,
-        dilation: int = 2,
-        torch_dtype: torch.dtype = torch.bfloat16,
-        device: torch.device = None,
-        **kwargs,
+    def __init__(
+    self,
+    bsz: int = 1,
+    dim: int = 896,
+    num_heads: int = 8,
+    num_layers: int = 12,
+    seq_len: int = 8192,
+    weight_tying: bool = False,
+    window_size: int = 1024,
+    vocab_size: int = 200064,
+    mlp_scale: int = 12,
+    bias: bool = False,
+    dropout: float = 0.0,
+    num_eigh: int = 24,
+    use_hankel_L: bool = False,
+    use_flash_fft: bool = True,
+    use_approx: bool = True,
+    use_attn: bool = True,
+    softcap: float = 50.0,
+    theta: float = 10_000.0,
+    use_alibi: bool = False,
+    dilation: int = 2,
+    torch_dtype: torch.dtype = torch.bfloat16,
+    device: torch.device = None,
+    **kwargs,
     ):
         super().__init__(**kwargs)
         self.bsz = bsz