config.json

{
  "architectures": [
    "Ernie4_5_VLMoeForConditionalGeneration"
  ],
  "dtype": "bfloat16",
  "image_end_token_id": 101305,
  "image_start_token_id": 101304,
  "image_token_id": 100295,
  "model_type": "ernie4_5_vl_moe",
  "text_config": {
    "hidden_act": "silu",
    "hidden_size": 2560,
    "initializer_range": 0.02,
    "intermediate_size": 12288,
    "max_position_embeddings": 131072,
    "mlp_layer_types": [
      "dense",
      "sparse",
      "sparse",
      "sparse"
    ],
    "model_type": "ernie4_5_vl_moe_text",
    "moe_intermediate_size": [
      1536,
      512
    ],
    "moe_k": 6,
    "moe_norm_min": 1e-12,
    "moe_num_experts": 64,
    "moe_num_shared_experts": 2,
    "num_attention_heads": 20,
    "num_hidden_layers": 4,
    "num_key_value_heads": 4,
    "output_router_logits": false,
    "rms_norm_eps": 1e-05,
    "rope_parameters": {
      "mrope_section": [
        22,
        22,
        20
      ],
      "rope_theta": 500000.0,
      "rope_type": "default"
    },
    "router_aux_loss_coef": 0.001,
    "use_bias": false,
    "use_cache": true,
    "vocab_size": 103424
  },
  "transformers_version": "5.0.0.dev0",
  "video_end_token_id": 101307,
  "video_start_token_id": 101306,
  "video_token_id": 103367,
  "vision_config": {
    "depth": 8,
    "hidden_act": "quick_gelu",
    "hidden_size": 1280,
    "in_channels": 3,
    "initializer_range": 0.02,
    "intermediate_size": 5120,
    "model_type": "ernie4_5_vl_moe_vision",
    "num_heads": 16,
    "patch_size": 14,
    "rms_norm_eps": 1e-06,
    "spatial_merge_size": 2,
    "temporal_merge_size": 2
  }
}