| { | |
| "architectures": [ | |
| "XCodec2Model" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "audio_channels": 1, | |
| "codebook_dim": 16, | |
| "codebook_size": 16384, | |
| "codec_decoder_hidden_size": 1024, | |
| "codec_encoder_hidden_size": 1024, | |
| "depth": 12, | |
| "dilations": [ | |
| 1, | |
| 3, | |
| 9 | |
| ], | |
| "head_dim": 64, | |
| "hidden_size": 1024, | |
| "hop_length": 320, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 4096, | |
| "max_position_embeddings": 4096, | |
| "model_type": "xcodec2", | |
| "num_attention_heads": 16, | |
| "num_hidden_layers": 12, | |
| "num_key_value_heads": 16, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 10000.0, | |
| "sampling_rate": 16000, | |
| "semantic_hidden_size": 1024, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.52.0.dev0", | |
| "up_ratios": [ | |
| 2, | |
| 2, | |
| 4, | |
| 4, | |
| 5 | |
| ], | |
| "use_vocos": true, | |
| "vq_commit_weight": 0.25, | |
| "vq_dim": 2048, | |
| "vq_full_commit_loss": false, | |
| "vq_num_quantizers": 1, | |
| "vq_weight_init": false | |
| } | |