{
  "architectures": [
    "SupraBrainForCausalLM"
  ],
  "attn_every": 4,
  "attn_head_dim": 64,
  "attn_num_heads": 6,
  "attn_num_kv_heads": 2,
  "attn_window": 256,
  "auto_map": {
    "AutoConfig": "modeling_suprabrain.SupraBrainConfig",
    "AutoModelForCausalLM": "modeling_suprabrain.SupraBrainForCausalLM"
  },
  "bos_token_id": 0,
  "ce_chunk": 1024,
  "compile_blocks": true,
  "dtype": "float32",
  "emb_shortcut": true,
  "eos_token_id": 2,
  "full_attn_layers": [
    19
  ],
  "gdn_conv_size": 4,
  "gdn_gate_rank": 32,
  "gdn_head_dim": 128,
  "gdn_num_heads": 3,
  "grad_ckpt": false,
  "hidden_size": 384,
  "initializer_range": 0.02,
  "intermediate_size": 1152,
  "logit_softcap": 0.0,
  "max_position_embeddings": 1024,
  "mlp_act": "rational",
  "model_type": "suprabrain",
  "norm_eps": 1e-05,
  "num_hidden_layers": 28,
  "pad_token_id": 1,
  "rope_theta": 10000.0,
  "surprise": true,
  "surprise_win": 16,
  "tie_word_embeddings": true,
  "transformers_version": "5.8.1",
  "unembed_rank": 32,
  "use_cache": false,
  "value_residual": true,
  "vocab_size": 23808,
  "z_loss": 0.0001
}