config fix

configuration_minitransformer.py

@@ -7,13 +7,13 @@ class MiniTransformerConfig(PretrainedConfig):
     def __init__(
         self,
         bsz: int = 1,
-        dim: int = 1536,
-        num_heads: int = 8,
-        num_layers: int = 26,
+        dim: int = 768,
+        num_heads: int = 24,
+        num_layers: int = 27,
         seq_len: int = 8192,
-        window_size: int = 1024,
+        window_size: int = 8192,
         vocab_size: int = 200064,
-        mlp_scale: int = 12,
+        mlp_scale: int = 4,
         bias: bool = False,
         dropout: float = 0.0,
         softcap: float = 50.0,