config.json

{
  "architectures": [
    "MossTTSNanoForCausalLM"
  ],
  "attn_implementation": "sdpa",
  "audio_assistant_slot_token_id": 9,
  "audio_codebook_sizes": [
    1024,
    1024,
    1024,
    1024,
    1024,
    1024,
    1024,
    1024,
    1024,
    1024,
    1024,
    1024,
    1024,
    1024,
    1024,
    1024
  ],
  "audio_end_token_id": 7,
  "audio_pad_token_id": 1024,
  "audio_start_token_id": 6,
  "audio_tokenizer_pretrained_name_or_path": "/workspace/MOSS-TTS-Nano/MOSS-TTS-Nano/models/MOSS-Audio-Tokenizer-Nano",
  "audio_tokenizer_sample_rate": 48000,
  "audio_tokenizer_type": "moss-audio-tokenizer-nano",
  "audio_user_slot_token_id": 8,
  "audio_vocab_size": 1024,
  "auto_map": {
    "AutoConfig": "configuration_moss_tts_nano.MossTTSNanoConfig",
    "AutoModel": "modeling_moss_tts_nano.MossTTSNanoForCausalLM",
    "AutoModelForCausalLM": "modeling_moss_tts_nano.MossTTSNanoForCausalLM"
  },
  "dtype": "float32",
  "gpt2_config": {
    "activation_function": "gelu_new",
    "attn_pdrop": 0.0,
    "bos_token_id": 1,
    "embd_pdrop": 0.0,
    "eos_token_id": 2,
    "initializer_range": 0.02,
    "layer_norm_epsilon": 1e-05,
    "model_type": "gpt2",
    "n_ctx": 32768,
    "n_embd": 768,
    "n_head": 12,
    "n_inner": 3072,
    "n_layer": 12,
    "n_positions": 32768,
    "pad_token_id": 3,
    "position_embedding_type": "rope",
    "reorder_and_upcast_attn": false,
    "resid_pdrop": 0.0,
    "rope_base": 10000.0,
    "scale_attn_by_inverse_layer_idx": false,
    "scale_attn_weights": true,
    "summary_activation": null,
    "summary_first_dropout": 0.1,
    "summary_proj_to_labels": true,
    "summary_type": "cls_index",
    "summary_use_proj": true,
    "use_cache": true,
    "vocab_size": 16384
  },
  "hidden_size": 768,
  "im_end_token_id": 5,
  "im_start_token_id": 4,
  "initializer_range": 0.02,
  "local_transformer_attn_implementation": "sdpa",
  "local_transformer_layers": 1,
  "max_position_embeddings": 32768,
  "model_architecture": "global_local_transformer",
  "model_type": "moss_tts_nano",
  "n_vq": 16,
  "pad_token_id": 3,
  "tokenizer_class": "MossTTSNanoSentencePieceTokenizer",
  "tokenizer_use_fast": false,
  "transformers_version": "4.57.1",
  "vocab_size": 16384
}