{
  "architectures": [
    "MossAudioTokenizerModel"
  ],
  "downsampling_ratios": [
    240,
    2,
    2,
    2
  ],
  "dtype": "float32",
  "hidden_sizes": [
    768,
    768,
    768,
    1280
  ],
  "input_hidden_sizes": [
    240,
    768,
    768,
    1280
  ],
  "intermediate_sizes": [
    3072,
    3072,
    3072,
    5120
  ],
  "layer_scale_init_value": 0.01,
  "max_position_embeddings": 2048,
  "model_type": "moss_audio_tokenizer",
  "num_attention_heads": [
    12,
    12,
    12,
    20
  ],
  "num_hidden_layers": [
    12,
    12,
    12,
    32
  ],
  "output_hidden_sizes": [
    384,
    384,
    640,
    768
  ],
  "quantizer_config": {
    "codebook_dim": 8,
    "codebook_size": 1024,
    "hidden_size": 512,
    "input_hidden_size": 768,
    "model_type": "",
    "n_codebooks": 32,
    "output_hidden_size": 768
  },
  "rope_parameters": {
    "rope_theta": 10000.0,
    "rope_type": "default"
  },
  "sampling_rate": 24000,
  "sliding_window_duration": 10,
  "transformers_version": "5.15.0.dev0",
  "use_cache": true
}