config.json

{
  "adapter_config": {
    "attention_dropout": 0.0,
    "float32_attention": true,
    "head_dim": 72,
    "hidden_act": "silu",
    "hidden_size": 1152,
    "image_feature_dropout": 0.0,
    "initializer_range": 0.02,
    "intermediate_size": 18944,
    "model_type": "",
    "num_attention_heads": 16,
    "num_key_value_heads": 16,
    "residual_dropout": 0.0,
    "text_hidden_size": 3584,
    "vit_layers": [
      -3,
      -9
    ]
  },
  "architectures": [
    "Molmo2ForConditionalGeneration"
  ],
  "auto_map": {
    "AutoConfig": "configuration_molmo2.Molmo2Config",
    "AutoModelForImageTextToText": "modeling_molmo2.Molmo2ForConditionalGeneration"
  },
  "image_patch_id": 152066,
  "initializer_range": 0.02,
  "llm_config": {
    "additional_vocab_size": 128,
    "attention_dropout": 0.0,
    "embedding_dropout": 0.0,
    "head_dim": 128,
    "hidden_act": "silu",
    "hidden_size": 3584,
    "initializer_range": 0.02,
    "intermediate_size": 18944,
    "layer_norm_eps": 1e-06,
    "max_position_embeddings": 4096,
    "model_type": "molmo2_llm",
    "norm_after": false,
    "num_attention_heads": 28,
    "num_hidden_layers": 28,
    "num_key_value_heads": 4,
    "qk_norm_type": "olmo",
    "qkv_bias": true,
    "residual_dropout": 0.0,
    "rope_scaling": null,
    "rope_theta": 1000000.0,
    "use_cache": true,
    "use_qk_norm": false,
    "vocab_size": 152064
  },
  "model_type": "molmo2",
  "tie_word_embeddings": false,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.52.3",
  "use_cache": true,
  "vit_config": {
    "attention_dropout": 0.0,
    "float32_attention": true,
    "head_dim": 72,
    "hidden_act": "gelu_pytorch_tanh",
    "hidden_size": 1152,
    "image_default_input_size": [
      378,
      378
    ],
    "image_num_pos": 729,
    "image_patch_size": 14,
    "initializer_range": 0.02,
    "intermediate_size": 4304,
    "layer_norm_eps": 1e-06,
    "model_type": "molmo2_vit",
    "num_attention_heads": 16,
    "num_hidden_layers": 27,
    "num_key_value_heads": 16,
    "residual_dropout": 0.0
  }
}