{
    "vocab_size": 50257,
    "embedding_dimension": 512,
    "num_heads": 8,
    "context_length": 256,
    "token_dropout": 0.03,
    "attn_dropout": 0.2,
    "ffn_dropout": 0.2,
    "qkv_bias": false,
    "num_layers": 8,
    "ff_hidden_dim": 1024,
    "rms_eps": 1e-06,
    "rms_bias": true,
    "theta_base": 10000.0,
    "num_kv_groups": 4,
    "num_experts": 4,
    "num_active_experts": 2,
    "moe_noise": true,
    "architectures": [
        "GQAGPT2"
    ],
    "model_type": "customGPT_pretrain"
}