{
  "architectures": [
    "JanusModel"
  ],
  "dtype": "float32",
  "image_token_id": 0,
  "model_type": "janus",
  "text_config": {
    "attention_bias": false,
    "attention_dropout": 0.0,
    "attention_probs_dropout_prob": 0.1,
    "bos_token_id": 100000,
    "eos_token_id": 100001,
    "head_dim": 8,
    "hidden_act": "gelu",
    "hidden_dropout_prob": 0.1,
    "hidden_size": 32,
    "id2label": {
      "0": "LABEL_0",
      "1": "LABEL_1",
      "2": "LABEL_2"
    },
    "initializer_range": 0.02,
    "intermediate_size": 37,
    "is_training": true,
    "label2id": {
      "LABEL_0": 0,
      "LABEL_1": 1,
      "LABEL_2": 2
    },
    "max_position_embeddings": 512,
    "mlp_bias": false,
    "model_type": "llama",
    "num_attention_heads": 4,
    "num_choices": 4,
    "num_hidden_layers": 2,
    "num_key_value_heads": 4,
    "pad_token_id": 100002,
    "pretraining_tp": 1,
    "rms_norm_eps": 1e-06,
    "rope_parameters": {
      "rope_theta": 10000.0,
      "rope_type": "default"
    },
    "seq_length": 7,
    "tie_word_embeddings": false,
    "type_sequence_label_size": 2,
    "type_vocab_size": 16,
    "use_cache": true,
    "use_input_mask": true,
    "use_labels": true,
    "use_token_type_ids": false,
    "vocab_size": 100590
  },
  "tie_word_embeddings": true,
  "transformers_version": "5.16.0.dev0",
  "vision_config": {
    "attention_bias": true,
    "attention_dropout": 0.1,
    "depth": 2,
    "dropout": 0.1,
    "hidden_act": "gelu",
    "hidden_dropout_rate": 0.0,
    "hidden_size": 32,
    "image_size": 20,
    "initializer_range": 0.02,
    "is_training": true,
    "layer_norm_eps": 1e-06,
    "mlp_ratio": 2,
    "model_type": "janus_vision_model",
    "num_attention_heads": 4,
    "num_channels": 3,
    "num_hidden_layers": 2,
    "num_image_tokens": 16,
    "num_key_value_heads": 1,
    "patch_size": 5,
    "projection_dim": 32,
    "projection_dropout": 0.0,
    "use_labels": true,
    "use_qk_norm": false,
    "vision_feature_layer": -1,
    "vision_feature_select_strategy": "default"
  },
  "vq_config": {
    "base_channels": 32,
    "channel_multiplier": [
      1,
      1
    ],
    "double_latent": false,
    "dropout": 0.0,
    "embed_dim": 12,
    "hidden_act": "gelu",
    "image_token_embed_dim": 32,
    "in_channels": 3,
    "initializer_range": 0.02,
    "latent_channels": 12,
    "model_type": "janus_vqgan",
    "num_embeddings": 12,
    "num_hidden_layers": 2,
    "num_patches": 4,
    "num_res_blocks": 2,
    "out_channels": 3,
    "projection_dim": 10
  }
}