{
  "architectures": [
    "ModernVBertForMaskedLM"
  ],
  "bos_token_id": 50281,
  "classifier_bias": false,
  "classifier_dropout": 0.0,
  "classifier_pooling": "cls",
  "dtype": "float32",
  "eos_token_id": 50282,
  "image_token_id": 98,
  "initializer_cutoff_factor": 2.0,
  "initializer_range": 0.02,
  "model_type": "modernvbert",
  "pad_token_id": 50283,
  "pixel_shuffle_factor": 2,
  "text_config": {
    "attention_bias": false,
    "attention_dropout": 0.1,
    "bos_token_id": 50281,
    "classifier_activation": "gelu",
    "classifier_bias": false,
    "classifier_dropout": 0.1,
    "classifier_pooling": "cls",
    "cls_token_id": 50281,
    "decoder_bias": true,
    "deterministic_flash_attn": false,
    "embedding_dropout": 0.1,
    "eos_token_id": 50282,
    "hidden_activation": "gelu",
    "hidden_size": 32,
    "initializer_cutoff_factor": 2.0,
    "initializer_range": 0.02,
    "intermediate_size": 64,
    "is_decoder": false,
    "layer_types": [
      "full_attention",
      "sliding_attention"
    ],
    "local_attention": 128,
    "max_position_embeddings": 512,
    "mlp_bias": false,
    "mlp_dropout": 0.1,
    "model_type": "modernbert",
    "norm_bias": false,
    "norm_eps": 1e-05,
    "num_attention_heads": 4,
    "num_hidden_layers": 2,
    "pad_token_id": 50283,
    "rope_parameters": {
      "full_attention": {
        "rope_theta": 160000.0,
        "rope_type": "default"
      },
      "sliding_attention": {
        "rope_theta": 10000.0,
        "rope_type": "default"
      }
    },
    "sep_token_id": 50282,
    "sparse_pred_ignore_index": -100,
    "sparse_prediction": false,
    "tie_word_embeddings": true,
    "type_vocab_size": 2,
    "vocab_size": 50408
  },
  "tie_word_embeddings": true,
  "transformers_version": "5.16.0.dev0",
  "vision_config": {
    "attention_dropout": 0.1,
    "dropout": 0.1,
    "hidden_act": "gelu_pytorch_tanh",
    "hidden_size": 64,
    "image_size": 16,
    "initializer_range": 0.02,
    "intermediate_size": 32,
    "layer_norm_eps": 1e-06,
    "model_type": "siglip_vision_model",
    "num_attention_heads": 4,
    "num_channels": 3,
    "num_hidden_layers": 2,
    "patch_size": 4
  },
  "vocab_size": 50408
}