config.json

{
  "architectures": [
    "DFlashDraftModel"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "auto_map": {
    "AutoModel": "dflash.DFlashDraftModel"
  },
  "block_size": 16,
  "bos_token_id": 0,
  "dflash_config": {
    "causal_head": true,
    "mask_token_id": 128796,
    "target_layer_ids": [
      1,
      9,
      17,
      25,
      33,
      41
    ]
  },
  "dtype": "bfloat16",
  "eos_token_id": 128007,
  "head_dim": 128,
  "hidden_act": "silu",
  "hidden_size": 4096,
  "initializer_range": 0.02,
  "intermediate_size": 7744,
  "layer_types": [
    "full_attention",
    "full_attention",
    "full_attention",
    "full_attention",
    "full_attention",
    "full_attention"
  ],
  "max_position_embeddings": 262144,
  "max_window_layers": 6,
  "model_type": "qwen3",
  "num_attention_heads": 32,
  "num_hidden_layers": 6,
  "num_key_value_heads": 4,
  "num_target_layers": 45,
  "pad_token_id": 2,
  "rms_norm_eps": 1e-05,
  "rope_parameters": {
    "rope_theta": 10000000,
    "rope_type": "default"
  },
  "sliding_window": null,
  "target_hidden_size": 4096,
  "tie_word_embeddings": false,
  "transformers_version": "5.8.1",
  "use_cache": true,
  "use_sliding_window": false,
  "vocab_size": 128896
}