config.json

{
  "activation_dropout": 0.0,
  "activation_function": "relu",
  "adaptor_dropout": 0.1,
  "adaptor_kernel_size": 8,
  "adaptor_stride": 8,
  "add_adapter": true,
  "architectures": [
    "SeamlessM4Tv2Model"
  ],
  "attention_dropout": 0.1,
  "bos_token_id": 2,
  "char_vocab_size": 10943,
  "conv_depthwise_kernel_size": 31,
  "decoder_attention_heads": 16,
  "decoder_ffn_dim": 8192,
  "decoder_layerdrop": 0.05,
  "decoder_layers": 24,
  "decoder_start_token_id": 3,
  "dropout": 0.1,
  "encoder_attention_heads": 16,
  "encoder_ffn_dim": 8192,
  "encoder_layerdrop": 0.05,
  "encoder_layers": 24,
  "eos_token_id": 3,
  "feature_projection_input_dim": 160,
  "hidden_size": 1024,
  "initializer_range": 0.02,
  "is_encoder_decoder": true,
  "lang_embed_dim": 256,
  "layer_norm_eps": 1e-05,
  "leaky_relu_slope": 0.1,
  "left_max_position_embeddings": 64,
  "max_new_tokens": 256,
  "max_position_embeddings": 4096,
  "model_type": "seamless_m4t_v2",
  "num_adapter_layers": 1,
  "num_attention_heads": 16,
  "num_hidden_layers": 24,
  "pad_token_id": 0,
  "position_embeddings_type": "relative_key",
  "resblock_dilation_sizes": [
    [
      1,
      3,
      5
    ],
    [
      1,
      3,
      5
    ],
    [
      1,
      3,
      5
    ]
  ],
  "resblock_kernel_sizes": [
    3,
    7,
    11
  ],
  "right_max_position_embeddings": 8,
  "sampling_rate": 16000,
  "scale_embedding": true,
  "speech_encoder_attention_heads": 16,
  "speech_encoder_chunk_size": 20000,
  "speech_encoder_dropout": 0.0,
  "speech_encoder_hidden_act": "swish",
  "speech_encoder_intermediate_size": 4096,
  "speech_encoder_layerdrop": 0.1,
  "speech_encoder_layers": 24,
  "speech_encoder_left_chunk_num": 128,
  "spkr_embed_dim": 256,
  "t2u_bos_token_id": 0,
  "t2u_decoder_attention_heads": 16,
  "t2u_decoder_ffn_dim": 8192,
  "t2u_decoder_layers": 6,
  "t2u_encoder_attention_heads": 16,
  "t2u_encoder_ffn_dim": 8192,
  "t2u_encoder_layers": 6,
  "t2u_eos_token_id": 2,
  "t2u_max_position_embeddings": 4096,
  "t2u_pad_token_id": 1,
  "t2u_variance_pred_dropout": 0.5,
  "t2u_variance_predictor_embed_dim": 1024,
  "t2u_variance_predictor_hidden_dim": 256,
  "t2u_variance_predictor_kernel_size": 3,
  "t2u_vocab_size": 10082,
  "torch_dtype": "float32",
  "transformers_version": "4.36.0.dev0",
  "unit_embed_dim": 1280,
  "unit_hifi_gan_vocab_size": 10000,
  "upsample_initial_channel": 512,
  "upsample_kernel_sizes": [
    11,
    8,
    8,
    4,
    4
  ],
  "upsample_rates": [
    5,
    4,
    4,
    2,
    2
  ],
  "use_cache": true,
  "var_pred_dropout": 0.5,
  "variance_predictor_kernel_size": 3,
  "vocab_size": 256102,
  "vocoder_num_langs": 36,
  "vocoder_num_spkrs": 200,
  "vocoder_offset": 4
}