{
  "architectures": [
    "Qwen3MoeForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "bos_token_id": 151643,
  "decoder_sparse_step": 1,
  "dtype": "bfloat16",
  "eos_token_id": 151645,
  "head_dim": 128,
  "hidden_act": "silu",
  "hidden_size": 2048,
  "initializer_range": 0.02,
  "intermediate_size": 6144,
  "max_position_embeddings": 262144,
  "max_window_layers": 48,
  "mlp_only_layers": [],
  "model_type": "qwen3_moe",
  "moe_intermediate_size": 768,
  "norm_topk_prob": true,
  "num_attention_heads": 32,
  "num_experts": 128,
  "num_experts_per_tok": 8,
  "num_hidden_layers": 48,
  "num_key_value_heads": 4,
  "output_router_logits": false,
  "quantization_config": {
    "bits": 4,
    "checkpoint_format": "gptq",
    "desc_act": false,
    "format": "gptq",
    "group_size": 128,
    "lm_head": false,
    "meta": {
      "act_group_aware": true,
      "auto_forward_data_parallel": true,
      "damp_auto_increment": 0.01,
      "damp_percent": 0.01,
      "failsafe": null,
      "gc_mode": "on_stage_end",
      "gptaq": false,
      "gptaq_alpha": 0.25,
      "moe_bypass_router": true,
      "mse": 0.0,
      "quantizer": [
        "gptqmodel:5.7.0-dev"
      ],
      "static_groups": false,
      "true_sequential": true,
      "uri": "https://github.com/modelcloud/gptqmodel",
      "wait_for_submodule_finalizers": false
    },
    "pack_dtype": "int32",
    "pack_impl": "cpu",
    "quant_method": "gptq",
    "sym": true
  },
  "rms_norm_eps": 1e-06,
  "rope_scaling": null,
  "rope_theta": 10000000,
  "router_aux_loss_coef": 0.001,
  "sliding_window": null,
  "tie_word_embeddings": false,
  "transformers_version": "4.57.3",
  "use_cache": true,
  "use_sliding_window": false,
  "vocab_size": 151936
}