{
  "architectures": [
    "DFlash2DraftModel"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "bos_token_id": null,
  "is_causal": false,
  "dflash_config": {
    "block_size": 8,
    "conv_group_size": 16,
    "conv_kernel_size": 2,
    "mask_token_id": 248070,
    "selector_rank": 256,
    "selector_top_k": 16,
    "target_layer_ids": [
      5,
      19,
      33,
      47,
      61
    ]
  },
  "dtype": "bfloat16",
  "eos_token_id": 248044,
  "head_dim": 128,
  "hidden_act": "silu",
  "hidden_size": 5120,
  "initializer_range": 0.02,
  "intermediate_size": 17408,
  "layer_types": [
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "sliding_attention",
    "sliding_attention"
  ],
  "max_position_embeddings": 262144,
  "max_window_layers": 5,
  "model_type": "qwen3",
  "num_attention_heads": 32,
  "num_hidden_layers": 5,
  "num_key_value_heads": 8,
  "num_target_layers": 64,
  "pad_token_id": 248044,
  "rms_norm_eps": 1e-06,
  "rope_parameters": {
    "rope_theta": 10000000,
    "rope_type": "default"
  },
  "sliding_window": 2048,
  "tie_word_embeddings": false,
  "transformers_version": "5.15.0",
  "use_cache": true,
  "use_sliding_window": true,
  "vocab_size": 248320,
  "quantization_config": {
    "activation_scheme": "dynamic",
    "fmt": "e4m3",
    "quant_method": "fp8",
    "modules_to_not_convert": [
      "candidate_selector.hidden_projection",
      "candidate_selector.predecessor_codebook",
      "candidate_selector.successor_codebook",
      "hidden_norm",
      "layers.0.attention_conv.base_kernel",
      "layers.0.attention_conv.kernel_projection",
      "layers.0.input_layernorm",
      "layers.0.mlp_conv.base_kernel",
      "layers.0.mlp_conv.kernel_projection",
      "layers.0.post_attention_layernorm",
      "layers.0.self_attn.k_norm",
      "layers.0.self_attn.q_norm",
      "layers.1.attention_conv.base_kernel",
      "layers.1.attention_conv.kernel_projection",
      "layers.1.input_layernorm",
      "layers.1.mlp_conv.base_kernel",
      "layers.1.mlp_conv.kernel_projection",
      "layers.1.post_attention_layernorm",
      "layers.1.self_attn.k_norm",
      "layers.1.self_attn.q_norm",
      "layers.2.attention_conv.base_kernel",
      "layers.2.attention_conv.kernel_projection",
      "layers.2.input_layernorm",
      "layers.2.mlp_conv.base_kernel",
      "layers.2.mlp_conv.kernel_projection",
      "layers.2.post_attention_layernorm",
      "layers.2.self_attn.k_norm",
      "layers.2.self_attn.q_norm",
      "layers.3.attention_conv.base_kernel",
      "layers.3.attention_conv.kernel_projection",
      "layers.3.input_layernorm",
      "layers.3.mlp_conv.base_kernel",
      "layers.3.mlp_conv.kernel_projection",
      "layers.3.post_attention_layernorm",
      "layers.3.self_attn.k_norm",
      "layers.3.self_attn.q_norm",
      "layers.4.attention_conv.base_kernel",
      "layers.4.attention_conv.kernel_projection",
      "layers.4.input_layernorm",
      "layers.4.mlp_conv.base_kernel",
      "layers.4.mlp_conv.kernel_projection",
      "layers.4.post_attention_layernorm",
      "layers.4.self_attn.k_norm",
      "layers.4.self_attn.q_norm",
      "norm"
    ],
    "weight_block_size": [
      128,
      128
    ]
  }
}