config.json

{
  "acoustic_tokenizer_chunk_size": 400,
  "acoustic_tokenizer_encoder_config": {
    "channels": 1,
    "depths": [
      1,
      1
    ],
    "downsampling_ratios": [
      2
    ],
    "ffn_expansion": 4,
    "hidden_act": "gelu",
    "hidden_size": 16,
    "initializer_range": 0.01,
    "kernel_size": 3,
    "layer_scale_init_value": 1e-06,
    "model_type": "vibevoice_acoustic_tokenizer_encoder",
    "n_filters": 4,
    "num_filters": 32,
    "rms_norm_eps": 1e-05,
    "vae_std": 0.625
  },
  "architectures": [
    "VibeVoiceAsrModel"
  ],
  "audio_bos_token_id": 151646,
  "audio_eos_token_id": 151647,
  "audio_token_id": 0,
  "dtype": "float32",
  "model_type": "vibevoice_asr",
  "semantic_tokenizer_encoder_config": {
    "channels": 1,
    "depths": [
      1,
      1
    ],
    "downsampling_ratios": [
      2
    ],
    "ffn_expansion": 4,
    "hidden_act": "gelu",
    "hidden_size": 32,
    "initializer_range": 0.01,
    "kernel_size": 3,
    "layer_scale_init_value": 1e-06,
    "model_type": "vibevoice_acoustic_tokenizer_encoder",
    "n_filters": 4,
    "num_filters": 32,
    "rms_norm_eps": 1e-05,
    "vae_std": 0.625
  },
  "text_config": {
    "attention_dropout": 0.0,
    "bos_token_id": null,
    "eos_token_id": null,
    "hidden_act": "silu",
    "hidden_size": 32,
    "initializer_range": 0.02,
    "intermediate_size": 36,
    "layer_types": [
      "full_attention",
      "full_attention"
    ],
    "max_position_embeddings": 200,
    "max_window_layers": 28,
    "model_type": "qwen2",
    "num_attention_heads": 4,
    "num_hidden_layers": 2,
    "num_key_value_heads": 4,
    "pad_token_id": 151643,
    "rms_norm_eps": 1e-06,
    "rope_parameters": {
      "rope_theta": 10000.0,
      "rope_type": "default"
    },
    "sliding_window": null,
    "tie_word_embeddings": false,
    "use_cache": true,
    "use_sliding_window": false,
    "vocab_size": 151665
  },
  "transformers_version": "5.16.0.dev0"
}