electra config: update vocab size & max pos for protein seq

Author: aditya0by0

configs/model/electra.yml

@@ -3,8 +3,8 @@ init_args:
   optimizer_kwargs:
     lr: 1e-3
   config:
-    vocab_size: 1400
-    max_position_embeddings: 1800
+    vocab_size: 31  # 21 amino acids (when n_gram=1) + 10 special tokens of LLM
+    max_position_embeddings: 1000  # max default sequence length for protein
     num_attention_heads: 8
     num_hidden_layers: 6
     type_vocab_size: 1