messing around with differing settings.

modeling_lsg_bert.py

@@ -62,13 +62,13 @@ class LSGBertConfig(BertConfig):
         base_model_prefix="lsg",
         block_size=128,
         sparse_block_size=128,
-        sparsity_factor=2,
         adaptive=True,
+        sparsity_factor=2,
         lsh_num_pre_rounds=1,
         mask_first_token=False,
         num_global_tokens=1,
         pool_with_global=True,
-        sparsity_type="norm",
+        sparsity_type="none",
         **kwargs
         ):
         """Constructs LSGBertConfig."""