config.json

{
  "architectures": [
    "H2OVLChatModel"
  ],
  "auto_map": {
    "AutoConfig": "configuration_h2ovl_chat.H2OVLChatConfig",
    "AutoModel": "modeling_h2ovl_chat.H2OVLChatModel",
    "AutoModelForCausalLM": "modeling_h2ovl_chat.H2OVLChatModel"
  },
  "downsample_ratio": 0.5,
  "dtype": "bfloat16",
  "dynamic_image_size": true,
  "force_image_size": 448,
  "llm_config": {
    "_name_or_path": "h2oai/h2o-danube2-1.8b-chat",
    "architectures": [
      "MistralForCausalLM"
    ],
    "attention_dropout": 0.0,
    "dtype": "bfloat16",
    "head_dim": null,
    "hidden_act": "silu",
    "hidden_size": 2560,
    "initializer_range": 0.02,
    "intermediate_size": 6912,
    "max_position_embeddings": 8192,
    "model_type": "mistral",
    "num_attention_heads": 32,
    "num_hidden_layers": 24,
    "num_key_value_heads": 8,
    "pad_token_id": 0,
    "rms_norm_eps": 1e-05,
    "rope_theta": 10000,
    "sliding_window": null,
    "use_bfloat16": true,
    "use_cache": true,
    "vocab_size": 32010
  },
  "max_dynamic_patch": 6,
  "min_dynamic_patch": 1,
  "model_type": "h2ovl_chat",
  "output_attentions": false,
  "pad2square": false,
  "ps_version": "v2",
  "select_layer": -1,
  "template": "h2ogpt2",
  "transformers_version": null,
  "use_backbone_lora": 0,
  "use_llm_lora": 0,
  "use_msac": true,
  "use_thumbnail": true,
  "vision_config": {
    "_name_or_path": "OpenGVLab/InternViT-300M-448px",
    "architectures": [
      "InternVisionModel"
    ],
    "attention_dropout": 0.0,
    "auto_map": {
      "AutoConfig": "OpenGVLab/InternViT-300M-448px--configuration_intern_vit.InternVisionConfig",
      "AutoModel": "OpenGVLab/InternViT-300M-448px--modeling_intern_vit.InternVisionModel"
    },
    "drop_path_rate": 0,
    "dropout": 0.0,
    "dtype": "bfloat16",
    "hidden_act": "gelu",
    "hidden_size": 1024,
    "image_size": 448,
    "initializer_factor": 1.0,
    "initializer_range": 0.02,
    "intermediate_size": 4096,
    "layer_norm_eps": 1e-06,
    "model_type": "intern_vit_6b",
    "norm_type": "layer_norm",
    "num_attention_heads": 16,
    "num_channels": 3,
    "num_hidden_layers": 24,
    "patch_size": 14,
    "qk_normalization": false,
    "qkv_bias": true,
    "use_flash_attn": true
  }
}