xtts2-gpt / config.json
mlinmg's picture
Update config.json
c98de0f verified
raw
history blame
2.25 kB
{
"_name_or_path": "AstraMindAI/xtts2-gpt",
"architectures": [
"XttsGPT"
],
"torch_dtype": "float32",
"auto_map": {
"AutoConfig": "AstraMindAI/xtts2-gpt--gpt_config.XTTSGPTConfig",
"AutoModelForCausalLM": "AstraMindAI/xtts2-gpt--xtts2_gpt_modeling.XttsGPT",
"AutoTokenizer": "AstraMindAI/xtts2-gpt--tokenizer.XTTSTokenizerFast"
},
"activation_function": "gelu",
"attn_pdrop": 0.1,
"audio_config": {
"fmax": 8000,
"fmin": 0,
"hop_length": 256,
"mel_channels": 80,
"mel_norms_file": null,
"n_fft": 1024,
"output_sample_rate": 24000,
"power": 1.0,
"sample_rate": 22050,
"win_length": 1024
},
"batch_size": 1,
"char_limits": {
"ar": 166,
"cs": 186,
"de": 253,
"en": 250,
"es": 239,
"fr": 273,
"hu": 224,
"it": 213,
"ja": 71,
"ko": 95,
"nl": 251,
"pl": 224,
"pt": 203,
"ru": 182,
"tr": 226,
"zh": 82
},
"checkpointing": false,
"code_stride_len": 1024,
"cond_chunk_len": 4,
"cond_len": 30,
"duration_const": 102400,
"embd_pdrop": 0.1,
"enable_redaction": false,
"hidden_size": 1024,
"kv_cache": true,
"label_smoothing": 0.0,
"languages": [
"en",
"es",
"fr",
"de",
"it",
"pt",
"pl",
"tr",
"ru",
"nl",
"cs",
"ar",
"zh-cn",
"hu",
"ko",
"ja",
"hi"
],
"layer_norm_epsilon": 1e-05,
"max_audio_tokens": 605,
"max_position_embeddings": 2048,
"max_prompt_tokens": 70,
"max_ref_len": 30,
"max_text_tokens": 402,
"model_type": "xtts_gpt",
"n_inner": null,
"num_attention_heads": 16,
"num_audio_tokens": 1026,
"num_chars": 255,
"num_hidden_layers": 30,
"number_text_tokens": 6681,
"perceiver_cond_length_compression": 256,
"reorder_and_upcast_attn": false,
"repetition_penalty": 5.0,
"resid_pdrop": 0.1,
"scale_attn_by_inverse_layer_idx": false,
"sound_norm_refs": false,
"start_audio_token": 1024,
"start_text_token": null,
"stop_audio_token": 1025,
"stop_text_token": null,
"temperature": 0.75,
"top_p": 0.85,
"train_solo_embeddings": false,
"transformers_version": "4.46.0",
"use_masking_gt_prompt_approach": true,
"use_perceiver_resampler": true,
"vocab_size": 256
}