microsoft/VibeVoice-ASR-BitNet

1{2  "_attn_implementation_autoset": false,3  "acoustic_tokenizer_config": {4    "causal": true,5    "channels": 1,6    "conv_bias": true,7    "conv_norm": "none",8    "corpus_normalize": 0.0,9    "decoder_depths": null,10    "decoder_n_filters": 32,11    "decoder_ratios": [12      8,13      5,14      5,15      4,16      2,17      218    ],19    "disable_last_norm": true,20    "dtype": "bfloat16",21    "encoder_depths": "3-3-3-3-3-3-8",22    "encoder_n_filters": 32,23    "encoder_ratios": [24      8,25      5,26      5,27      4,28      2,29      230    ],31    "fix_std": 0.5,32    "layer_scale_init_value": 1e-06,33    "layernorm": "RMSNorm",34    "layernorm_elementwise_affine": true,35    "layernorm_eps": 1e-05,36    "mixer_layer": "depthwise_conv",37    "model_type": "vibevoice_acoustic_tokenizer",38    "pad_mode": "constant",39    "std_dist_type": "gaussian",40    "vae_dim": 64,41    "weight_init_value": 0.0142  },43  "acoustic_vae_dim": 64,44  "architectures": [45    "VibeVoiceForASRTraining"46  ],47  "decoder_config": {48    "attention_dropout": 0.0,49    "dtype": "bfloat16",50    "hidden_act": "silu",51    "hidden_size": 1536,52    "initializer_range": 0.02,53    "intermediate_size": 8960,54    "layer_types": [55      "full_attention",56      "full_attention",57      "full_attention",58      "full_attention",59      "full_attention",60      "full_attention",61      "full_attention",62      "full_attention",63      "full_attention",64      "full_attention",65      "full_attention",66      "full_attention",67      "full_attention",68      "full_attention",69      "full_attention",70      "full_attention",71      "full_attention",72      "full_attention",73      "full_attention",74      "full_attention",75      "full_attention",76      "full_attention",77      "full_attention",78      "full_attention",79      "full_attention",80      "full_attention",81      "full_attention",82      "full_attention"83    ],84    "max_position_embeddings": 65536,85    "max_window_layers": 28,86    "model_type": "qwen2",87    "num_attention_heads": 12,88    "num_hidden_layers": 28,89    "num_key_value_heads": 2,90    "rms_norm_eps": 1e-06,91    "rope_scaling": null,92    "rope_theta": 1000000.0,93    "sliding_window": null,94    "tie_word_embeddings": true,95    "use_cache": true,96    "use_sliding_window": false,97    "vocab_size": 15193698  },99  "diffusion_head_config": {100    "ddpm_batch_mul": 4,101    "ddpm_beta_schedule": "cosine",102    "ddpm_num_inference_steps": 20,103    "ddpm_num_steps": 1000,104    "diffusion_type": "ddpm",105    "head_ffn_ratio": 3.0,106    "head_layers": 4,107    "hidden_size": 1536,108    "latent_size": 64,109    "model_type": "vibepod_diffusion_head",110    "prediction_type": "v_prediction",111    "rms_norm_eps": 1e-05,112    "speech_vae_dim": 64113  },114  "dtype": "float32",115  "model_type": "vibevoice",116  "semantic_tokenizer_config": {117    "causal": true,118    "channels": 1,119    "conv_bias": true,120    "conv_norm": "none",121    "corpus_normalize": 0.0,122    "disable_last_norm": true,123    "dtype": "bfloat16",124    "encoder_depths": "3-3-3-3-3-3-8",125    "encoder_n_filters": 32,126    "encoder_ratios": [127      8,128      5,129      5,130      4,131      2,132      2133    ],134    "fix_std": 0,135    "layer_scale_init_value": 1e-06,136    "layernorm": "RMSNorm",137    "layernorm_elementwise_affine": true,138    "layernorm_eps": 1e-05,139    "mixer_layer": "depthwise_conv",140    "model_type": "vibevoice_semantic_tokenizer",141    "pad_mode": "constant",142    "std_dist_type": "none",143    "vae_dim": 128,144    "weight_init_value": 0.01145  },146  "semantic_vae_dim": 128,147  "transformers_version": "4.57.6"148}149