config.json

{
  "arch_specifier": "no-align+fused-gelu-mlp",
  "architectures": [
    "PrismaticForConditionalGeneration"
  ],
  "auto_map": {
    "AutoConfig": "configuration_prismatic.PrismaticConfig",
    "AutoModelForVision2Seq": "modeling_prismatic.PrismaticForConditionalGeneration"
  },
  "hf_llm_id": "meta-llama/Llama-3.2-1B",
  "image_resize_strategy": "letterbox",
  "image_sizes": [
    224,
    224
  ],
  "llm_backbone_id": "llama3.2-1b",
  "llm_max_length": 2048,
  "model_type": "prismatic",
  "output_projector_states": false,
  "pad_to_multiple_of": 64,
  "pad_token_id": 128256,
  "text_config": {
    "architectures": [
      "LlamaForCausalLM"
    ],
    "bos_token_id": 128000,
    "eos_token_id": 128001,
    "head_dim": 64,
    "hidden_size": 2048,
    "intermediate_size": 8192,
    "max_position_embeddings": 131072,
    "model_type": "llama",
    "num_hidden_layers": 16,
    "num_key_value_heads": 8,
    "pad_token_id": 128256,
    "rms_norm_eps": 1e-05,
    "rope_theta": 500000.0,
    "tie_word_embeddings": true,
    "torch_dtype": "bfloat16",
    "vocab_size": 128320
  },
  "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": "4.45.1",
  "use_fused_vision_backbone": true,
  "vision_backbone_id": "dinosiglip-vit-so-224px"
}