net_config.pytorch.json

{
    "block_size": "4_4_4",
    "d_embed": 768,
    "d_head": 64,
    "d_inner": 3072,
    "d_model": 768,
    "dropact": 0.0,
    "dropatt": 0.1,
    "dropout": 0.1,
    "n_head": 12,
    "pool_q_only": true,
    "pooling_size": 2,
    "pooling_type": "mean",
    "separate_cls": true,
    "vocab_size": 50000
}