{
  "architectures": [
    "Qwen3MoeForCausalLM"
  ],
  "model_type": "qwen3_moe",
  "vocab_size": 151936,
  "hidden_size": 2048,
  "num_attention_heads": 32,
  "num_hidden_layers": 8,
  "head_dim": 128,
  "num_key_value_heads": 8,
  "rope_theta": 10000000.0,
  "max_position_embeddings": 262144,
  "rms_norm_eps": 1e-06,
  "hidden_act": "silu",
  "tie_word_embeddings": false,
  "use_cache": true,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.53.3",
  "attention_bias": false,
  "attention_dropout": 0.0,
  "qk_norm": true,
  "sliding_window": null,
  "use_sliding_window": false,
  "num_experts": 16,
  "num_experts_per_tok": 8,
  "moe_intermediate_size": 768,
  "intermediate_size": 768,
  "decoder_sparse_step": 1,
  "norm_topk_prob": false,
  "output_router_logits": false,
  "router_aux_loss_coef": 0.001
}