config.json

{
    "model_type": "OptimizedSimpleTransformer",
    "vocab_size": 50257,
    "num_layers": 2,
    "d_model": 128,
    "n_heads": 4,
    "d_head": 32,
    "block_size": 32,
    "btree_order": 5,
    "dropout": 0.3
}