{
  "architectures": [
    "Ovis2ForConditionalGeneration"
  ],
  "dtype": "float32",
  "hidden_size": 64,
  "image_token_id": 1,
  "model_type": "ovis2",
  "text_config": {
    "attention_dropout": 0.0,
    "bos_token_id": null,
    "eos_token_id": null,
    "hidden_act": "gelu",
    "hidden_size": 64,
    "id2label": {
      "0": "LABEL_0",
      "1": "LABEL_1",
      "2": "LABEL_2"
    },
    "initializer_range": 0.02,
    "intermediate_size": 54,
    "is_training": true,
    "label2id": {
      "LABEL_0": 0,
      "LABEL_1": 1,
      "LABEL_2": 2
    },
    "layer_types": [
      "full_attention",
      "full_attention"
    ],
    "max_position_embeddings": 580,
    "max_window_layers": 28,
    "model_type": "qwen2",
    "num_attention_heads": 4,
    "num_hidden_layers": 2,
    "num_key_value_heads": 4,
    "pad_token_id": 151643,
    "rms_norm_eps": 1e-06,
    "rope_parameters": {
      "rope_theta": 10000.0,
      "rope_type": "default"
    },
    "seq_length": 7,
    "sliding_window": null,
    "tie_word_embeddings": false,
    "use_cache": true,
    "use_labels": true,
    "use_sliding_window": false,
    "vocab_size": 151671
  },
  "tie_word_embeddings": true,
  "transformers_version": "5.16.0.dev0",
  "vision_config": {
    "attention_dropout": 0.0,
    "hidden_act": "silu",
    "hidden_size": 64,
    "hidden_stride": 2,
    "image_size": 32,
    "initializer_range": 0.02,
    "intermediate_size": 54,
    "mlp_bias": false,
    "model_type": "",
    "num_attention_heads": 4,
    "num_channels": 3,
    "num_hidden_layers": 2,
    "num_visual_indicator_tokens": 5,
    "patch_size": 8,
    "qkv_bias": false,
    "rms_norm_eps": 1e-05,
    "tokenize_function": "softmax",
    "vocab_size": 99
  },
  "visual_indicator_token_ids": [],
  "vocab_size": 151671
}