diminished block size and sparse_block_size to deal with leftovers

modeling_lsg_bert.py

@@ -60,8 +60,8 @@ class LSGBertConfig(BertConfig):
     def __init__(
         self,
         base_model_prefix="lsg",
-        block_size=128,
-        sparse_block_size=128,
+        block_size=64,
+        sparse_block_size=64,
         sparsity_factor=2,
         adaptive=True,
         lsh_num_pre_rounds=1,