{
  "_name_or_path": "TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T",
  "architectures": [
    "DetikzifyForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "bos_token_id": 1,
  "concat_patches": 2,
  "eos_token_id": 2,
  "feature_layer": 26,
  "hidden_act": "silu",
  "hidden_size": 2048,
  "initializer_range": 0.02,
  "intermediate_size": 5632,
  "max_position_embeddings": 2048,
  "mm_hidden_size": 2304,
  "model_type": "detikzify",
  "num_attention_heads": 32,
  "num_hidden_layers": 22,
  "num_key_value_heads": 4,
  "num_patches": 364,
  "pad_token_id": 32000,
  "patch_token_id": 1,
  "pretraining_tp": 1,
  "rms_norm_eps": 1e-05,
  "rope_scaling": null,
  "rope_theta": 10000.0,
  "tie_word_embeddings": false,
  "torch_dtype": "float32",
  "transformers_version": "4.36.2",
  "use_cache": false,
  "use_mm_proj": true,
  "vision_config": {
    "architecture": "vit_so400m_patch14_siglip_384",
    "classifier": "head",
    "crop_mode": "center",
    "crop_pct": 0.9,
    "custom_load": false,
    "first_conv": "patch_embed.proj",
    "fixed_input_size": true,
    "hf_hub_filename": "open_clip_pytorch_model.bin",
    "hf_hub_id": "timm/ViT-SO400M-14-SigLIP-384",
    "input_size": [
      3,
      384,
      384
    ],
    "interpolation": "bicubic",
    "mean": [
      0.5,
      0.5,
      0.5
    ],
    "num_classes": 0,
    "pool_size": null,
    "std": [
      0.5,
      0.5,
      0.5
    ],
    "tag": "webli",
    "url": ""
  },
  "vision_tower": "vit_so400m_patch14_siglip_384.webli",
  "vocab_size": 32008
}