config.json

{
  "architectures": [
    "JanusForConditionalGeneration"
  ],
  "dtype": "bfloat16",
  "image_token_id": 100581,
  "initializer_range": 0.02,
  "model_type": "janus",
  "text_config": {
    "attention_bias": false,
    "attention_dropout": 0.0,
    "bos_token_id": 100000,
    "dtype": "bfloat16",
    "eos_token_id": 100001,
    "head_dim": 128,
    "hidden_act": "silu",
    "hidden_size": 2048,
    "initializer_range": 0.02,
    "intermediate_size": 5632,
    "max_position_embeddings": 16384,
    "mlp_bias": false,
    "model_type": "llama",
    "num_attention_heads": 16,
    "num_hidden_layers": 1,
    "num_key_value_heads": 16,
    "pad_token_id": 100002,
    "pretraining_tp": 1,
    "rms_norm_eps": 1e-06,
    "rope_scaling": null,
    "rope_theta": 10000.0,
    "use_cache": true,
    "vocab_size": 102400
  },
  "transformers_version": "4.56.0",
  "vision_config": {
    "attention_bias": true,
    "attention_dropout": 0.0,
    "depth": 2,
    "dtype": "bfloat16",
    "hidden_act": "gelu",
    "hidden_dropout_rate": 0.0,
    "hidden_size": 1024,
    "image_size": 384,
    "initializer_range": 0.02,
    "layer_norm_eps": 1e-06,
    "mlp_ratio": 4.0,
    "model_type": "janus_vision_model",
    "num_attention_heads": 16,
    "num_channels": 3,
    "num_hidden_layers": 1,
    "num_image_tokens": 576,
    "patch_size": 16,
    "projection_dim": 2048,
    "projection_dropout": 0.0,
    "use_qk_norm": false
  },
  "vq_config": {
    "base_channels": 128,
    "channel_multiplier": [
      1,
      1,
      2,
      2,
      4
    ],
    "depth": 2,
    "double_latent": false,
    "dropout": 0.0,
    "dtype": "bfloat16",
    "embed_dim": 8,
    "hidden_act": "gelu",
    "image_token_embed_dim": 2048,
    "in_channels": 3,
    "initializer_range": 0.02,
    "latent_channels": 256,
    "model_type": "janus_vqgan",
    "num_embeddings": 16384,
    "num_hidden_layers": 2,
    "num_patches": 24,
    "num_res_blocks": 2,
    "out_channels": 3,
    "projection_dim": 2048
  }
}