{
  "architectures": [
    "QMoDForCausalLM"
  ],
  "attention_dropout": 0.0,
  "attention_scalar_gate": true,
  "bos_token_id": 1,
  "dtype": "float32",
  "eos_token_id": 2,
  "gate_multiplier": 2.0,
  "head_dim": 64,
  "hidden_size": 512,
  "intermediate_size": 1792,
  "max_position_embeddings": 2048,
  "mlp_scalar_gate": true,
  "mod_capacity": 0.125,
  "mod_intermediate_size": 1797,
  "mod_layers": [
    1,
    3,
    5,
    7,
    9
  ],
  "model_type": "q_mod",
  "nope_every_n": null,
  "nope_layers": [],
  "num_attention_heads": 8,
  "num_hidden_layers": 10,
  "num_key_value_heads": 2,
  "pad_token_id": 0,
  "qk_norm": true,
  "rms_norm_eps": 1e-05,
  "rope_theta": 10000.0,
  "router_aux_coef": 0.01,
  "tie_word_embeddings": true,
  "transformers_version": "5.17.0",
  "vocab_size": 32768,
  "auto_map": {
    "AutoConfig": "configuration_qmod.QMoDConfig",
    "AutoModelForCausalLM": "modeling_qmod.QMoDForCausalLM"
  },
  "_name_or_path": "d0rj/q-mod-51M-base"
}