{
  "audio_token_id": 131085,
  "audio_token_offset": 262344,
  "audio_tokenizer_config": {
    "adanorm_num_embeddings": 4,
    "architectures": [
      "WavTokenizerModel"
    ],
    "audio_channels": 1,
    "codebook_dim": 512,
    "codebook_size": 4096,
    "compress": 2,
    "decoder_attention_num_groups": 32,
    "decoder_hidden_size": 768,
    "decoder_intermediate_size": 2304,
    "decoder_num_layers": 12,
    "dilation_growth_rate": 2,
    "dtype": "float32",
    "hidden_size": 512,
    "kernel_size": 7,
    "last_kernel_size": 7,
    "model_type": "wavtokenizer",
    "norm_eps": 1e-06,
    "norm_type": "weight_norm",
    "num_filters": 32,
    "num_lstm_layers": 2,
    "num_residual_layers": 1,
    "pad_mode": "reflect",
    "residual_kernel_size": 3,
    "sampling_rate": 24000,
    "upsampling_ratios": [
      6,
      5,
      5,
      4
    ],
    "use_causal_conv": false,
    "use_conv_shortcut": true
  },
  "image_token_id": 131079,
  "image_token_offset": 131272,
  "model_type": "apertus1p5",
  "text_config": {
    "attention_bias": false,
    "attention_dropout": 0.0,
    "bos_token_id": 1,
    "dtype": "bfloat16",
    "eos_token_id": [
      2,
      68,
      72
    ],
    "hidden_act": "xielu",
    "hidden_dropout": 0.0,
    "hidden_size": 8192,
    "initializer_range": 0.02,
    "intermediate_size": 43008,
    "max_position_embeddings": 262144,
    "mlp_bias": false,
    "model_type": "apertus1p5_text",
    "num_attention_heads": 64,
    "num_hidden_layers": 80,
    "num_key_value_heads": 8,
    "output_vocab_size": 131072,
    "pad_token_id": 3,
    "post_norm": false,
    "qk_norm": true,
    "rms_norm_eps": 1e-05,
    "rope_parameters": {
      "factor": 32.0,
      "high_freq_factor": 4.0,
      "low_freq_factor": 1.0,
      "original_max_position_embeddings": 8192,
      "rope_theta": 4000000,
      "rope_type": "llama3"
    },
    "tie_word_embeddings": false,
    "use_cache": true,
    "vocab_size": 266752
  },
  "tie_word_embeddings": false,
  "transformers_version": "5.14.0.dev0",
  "vision_tokenizer_config": {
    "architectures": [
      "Apertus1p5VisionTokenizerModel"
    ],
    "attn_resolutions": [
      16
    ],
    "base_channels": 256,
    "channel_multiplier": [
      1,
      1,
      2,
      2,
      4
    ],
    "codebook_size": 131072,
    "dropout": 0.0,
    "dtype": "float32",
    "embed_dim": 256,
    "in_channels": 3,
    "latent_channels": 256,
    "model_type": "apertus1p5_vision_tokenizer",
    "num_res_blocks": 4,
    "resolution": 256
  },
  "architectures": [
    "Apertus1p5ForConditionalGeneration"
  ]
}