config.json

{
  "arch_specifier": "no-align+gelu-mlp",
  "architectures": [
    "PrismaticForConditionalGeneration"
  ],
  "hf_llm_id": "meta-llama/Meta-Llama-3.1-8B",
  "image_resize_strategy": "letterbox",
  "image_sizes": [
    224,
    224
  ],
  "llm_backbone_id": "llama3-1-8b-pure",
  "llm_max_length": 2048,
  "model_type": "prismatic",
  "output_projector_states": false,
  "pad_to_multiple_of": 64,
  "pad_token_id": 0,
  "padding_idx": 0,
  "text_config": {
    "model_type": "llama",
    "pad_token_id": 128256,
    "torch_dtype": "bfloat16",
    "vocab_size": 128064
  },
  "timm_model_ids": [
    "vit_large_patch14_reg4_dinov2.lvd142m",
    "vit_so400m_patch14_siglip_224"
  ],
  "timm_override_act_layers": [
    null,
    null
  ],
  "torch_dtype": "bfloat16",
  "transformers_version": "4.43.3",
  "use_fused_vision_backbone": true,
  "vision_backbone_id": "dinosiglip-vit-so-224px",
  "vocab_size": 32001
}