{
  "architectures": [
    "ExaoneMoeForCausalLM"
  ],
  "attention_dropout": 0.0,
  "bos_token_id": 1,
  "dtype": "bfloat16",
  "eos_token_id": 53,
  "first_k_dense_replace": 1,
  "head_dim": 128,
  "hidden_act": "silu",
  "hidden_size": 6144,
  "initializer_range": 0.02,
  "intermediate_size": 18432,
  "is_moe_layer": [
    false,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true
  ],
  "layer_types": [
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "full_attention"
  ],
  "max_position_embeddings": 262144,
  "mlp_layer_types": [
    "dense",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse"
  ],
  "model_type": "exaone_moe",
  "moe_intermediate_size": 2048,
  "mtp_layer_types": [
    "full_attention"
  ],
  "mtp_sliding_windows": [
    0
  ],
  "n_group": 1,
  "norm_topk_prob": true,
  "num_attention_heads": 64,
  "num_experts": 128,
  "num_experts_per_tok": 8,
  "num_hidden_layers": 48,
  "num_key_value_heads": 8,
  "num_nextn_predict_layers": 1,
  "num_shared_experts": 1,
  "pad_token_id": 0,
  "rms_norm_eps": 1e-05,
  "rope_parameters": {
    "rope_theta": 1000000,
    "rope_type": "default"
  },
  "routed_scaling_factor": 2.5,
  "scoring_func": "sigmoid",
  "sliding_window": 128,
  "sliding_window_pattern": "LLLG",
  "sliding_windows": [
    128,
    128,
    128,
    0,
    128,
    128,
    128,
    0,
    128,
    128,
    128,
    0,
    128,
    128,
    128,
    0,
    128,
    128,
    128,
    0,
    128,
    128,
    128,
    0,
    128,
    128,
    128,
    0,
    128,
    128,
    128,
    0,
    128,
    128,
    128,
    0,
    128,
    128,
    128,
    0,
    128,
    128,
    128,
    0,
    128,
    128,
    128,
    0
  ],
  "tie_word_embeddings": false,
  "tokenizer_class": "GPT2Tokenizer",
  "topk_group": 1,
  "transformers_version": "5.1.0",
  "use_cache": true,
  "vocab_size": 153600
}