config.json

{
  "architectures": [
    "Granite4VisionForConditionalGeneration"
  ],
  "auto_map": {
    "AutoConfig": "configuration.Granite4VisionConfig",
    "AutoModel": "modeling.Granite4VisionForConditionalGeneration",
    "AutoModelForImageTextToText": "modeling.Granite4VisionForConditionalGeneration",
    "AutoModelForVision2Seq": "modeling.Granite4VisionForConditionalGeneration",
    "AutoProcessor": "processing.Granite4VisionProcessor"
  },
  "deepstack_layer_map": [
    [
      -4,
      3
    ],
    [
      -3,
      2
    ],
    [
      -2,
      1
    ],
    [
      -1,
      0
    ]
  ],
  "downsample_rate": "4/8",
  "dtype": "float32",
  "image_grid_pinpoints": [
    [
      384,
      384
    ],
    [
      384,
      768
    ],
    [
      384,
      1152
    ],
    [
      384,
      1536
    ],
    [
      384,
      1920
    ],
    [
      384,
      2304
    ],
    [
      384,
      2688
    ],
    [
      384,
      3072
    ],
    [
      384,
      3456
    ],
    [
      384,
      3840
    ],
    [
      768,
      384
    ],
    [
      768,
      768
    ],
    [
      768,
      1152
    ],
    [
      768,
      1536
    ],
    [
      768,
      1920
    ],
    [
      1152,
      384
    ],
    [
      1152,
      768
    ],
    [
      1152,
      1152
    ],
    [
      1536,
      384
    ],
    [
      1536,
      768
    ],
    [
      1920,
      384
    ],
    [
      1920,
      768
    ],
    [
      2304,
      384
    ],
    [
      2688,
      384
    ],
    [
      3072,
      384
    ],
    [
      3456,
      384
    ],
    [
      3840,
      384
    ]
  ],
  "image_seq_length": 576,
  "image_token_index": 100266,
  "initializer_range": 0.02,
  "model_type": "granite4_vision",
  "projector_dropout": 0.1,
  "projector_hidden_act": "gelu",
  "qformer_config": {
    "attention_probs_dropout_prob": 0.1,
    "cross_attention_frequency": 1,
    "dtype": "float32",
    "encoder_hidden_size": 384,
    "hidden_act": "gelu",
    "hidden_dropout_prob": 0.1,
    "hidden_size": 384,
    "initializer_range": 0.02,
    "intermediate_size": 768,
    "layer_norm_eps": 1e-12,
    "max_position_embeddings": 2048,
    "model_type": "blip_2_qformer",
    "num_attention_heads": 4,
    "num_hidden_layers": 1,
    "pad_token_id": 0,
    "use_qformer_text_input": false,
    "vocab_size": 30522
  },
  "spatial_stride": 2,
  "spatial_target_layers": [
    0,
    1,
    2,
    3
  ],
  "spatial_vision_layer": -1,
  "text_config": {
    "architectures": [
      "GraniteForCausalLM"
    ],
    "attention_bias": false,
    "attention_dropout": 0.0,
    "attention_multiplier": 0.015625,
    "bos_token_id": 100257,
    "dtype": "float32",
    "embedding_multiplier": 12.0,
    "eos_token_id": 100257,
    "hidden_act": "silu",
    "hidden_size": 1024,
    "initializer_range": 0.1,
    "intermediate_size": 2048,
    "logits_scaling": 10.0,
    "max_position_embeddings": 131072,
    "mlp_bias": false,
    "model_type": "granite",
    "num_attention_heads": 8,
    "num_hidden_layers": 4,
    "num_key_value_heads": 2,
    "pad_token_id": 100256,
    "residual_multiplier": 0.22,
    "rms_norm_eps": 1e-05,
    "rope_parameters": {
      "rope_theta": 10000000,
      "rope_type": "default"
    },
    "tie_word_embeddings": true,
    "use_cache": true,
    "vocab_size": 100352
  },
  "tie_word_embeddings": true,
  "transformers_version": "5.14.1",
  "use_cache": true,
  "use_image_newline_parameter": true,
  "use_spatial_sampling": true,
  "vision_config": {
    "attention_dropout": 0.0,
    "dtype": "float32",
    "hidden_act": "gelu_pytorch_tanh",
    "hidden_size": 384,
    "image_size": 384,
    "intermediate_size": 768,
    "layer_norm_eps": 1e-06,
    "model_type": "siglip_vision_model",
    "num_attention_heads": 4,
    "num_channels": 3,
    "num_hidden_layers": 4,
    "patch_size": 16
  },
  "vision_feature_layer": -2,
  "vision_feature_select_strategy": "full"
}