config.json

{
  "architectures": [
    "Sarashina2VisionForCausalLM"
  ],
  "auto_map": {
    "AutoConfig": "configuration_sarashina2_vision.Sarashina2VisionConfig",
    "AutoModelForCausalLM": "modeling_sarashina2_vision.Sarashina2VisionForCausalLM"
  },
  "bos_token_id": 1,
  "dtype": "bfloat16",
  "eos_token_id": 2,
  "hidden_act": "silu",
  "image_token_id": 14,
  "model_type": "sarashina2_vision",
  "pad_token_id": 3,
  "rope_scaling": {
    "mrope_interleaved": true,
    "mrope_section": [
      20,
      30,
      30
    ],
    "rope_type": "default",
    "spatial_reset": true
  },
  "text_config": {
    "_name_or_path": "sbintuitions/sarashina2.2-3b-instruct-v0.1",
    "architectures": [
      "LlamaForCausalLM"
    ],
    "attention_bias": false,
    "attention_dropout": 0.0,
    "dtype": "bfloat16",
    "head_dim": 160,
    "hidden_act": "silu",
    "hidden_size": 2560,
    "initializer_range": 0.02,
    "intermediate_size": 8960,
    "max_position_embeddings": 16384,
    "mlp_bias": false,
    "model_type": "llama",
    "num_attention_heads": 16,
    "num_hidden_layers": 32,
    "num_key_value_heads": 8,
    "pretraining_tp": 1,
    "rms_norm_eps": 1e-05,
    "rope_scaling": null,
    "rope_theta": 500000,
    "use_cache": true,
    "vocab_size": 102400
  },
  "transformers_version": "4.57.1",
  "video_token_id": 102399,
  "vision_config": {
    "deepstack_visual_indices": [
      6,
      13,
      20
    ],
    "depth": 27,
    "embed_dim": 1152,
    "hidden_act": "gelu_pytorch_tanh",
    "hidden_size": 2560,
    "in_channels": 3,
    "initializer_range": 0.02,
    "mlp_ratio": 3.7362,
    "model_type": "qwen2_vl",
    "num_heads": 16,
    "patch_size": 14,
    "spatial_merge_size": 2,
    "temporal_patch_size": 2
  },
  "vision_end_token_id": 102398,
  "vision_start_token_id": 102397,
  "vocab_size": 102400
}