| allow_cache: true |
| batch_size: 1 |
| config: ./ttsexamples/tacotron2/conf/tacotron2.v1.yaml |
| dev_dir: /content/dump_ljspeech/valid |
| end_ratio_value: 0.0 |
| eval_interval_steps: 50 |
| format: npy |
| gradient_accumulation_steps: 1 |
| hop_size: 256 |
| is_shuffle: true |
| log_interval_steps: 20 |
| max_char_length: 128 |
| max_mel_length: 257 |
| mel_length_threshold: 32 |
| mixed_precision: false |
| model_type: tacotron2 |
| num_save_intermediate_results: 1 |
| optimizer_params: {decay_steps: 150000, end_learning_rate: 1.0e-05, initial_learning_rate: 0.001, |
| warmup_proportion: 0.02, weight_decay: 0.001} |
| outdir: ./ttsexamples/tacotron2/exp/train.tacotron2.v1/ |
| pretrained: '' |
| remove_short_samples: true |
| resume: '' |
| save_interval_steps: 20 |
| schedule_decay_steps: 50000 |
| start_ratio_value: 0.5 |
| start_schedule_teacher_forcing: 200001 |
| tacotron2_params: {attention_dim: 128, attention_filters: 32, attention_kernel: 31, |
| attention_type: lsa, dataset: ljspeech, decoder_lstm_units: 1024, embedding_dropout_prob: 0.1, |
| embedding_hidden_size: 512, encoder_conv_activation: relu, encoder_conv_dropout_rate: 0.5, |
| encoder_conv_filters: 512, encoder_conv_kernel_sizes: 5, encoder_lstm_units: 256, |
| initializer_range: 0.02, n_conv_encoder: 5, n_conv_postnet: 5, n_lstm_decoder: 1, |
| n_mels: 80, n_prenet_layers: 2, n_speakers: 1, postnet_conv_filters: 512, postnet_conv_kernel_sizes: 5, |
| postnet_dropout_rate: 0.1, prenet_activation: relu, prenet_dropout_rate: 0.5, prenet_units: 256, |
| reduction_factor: 1} |
| train_dir: /content/dump_ljspeech/train |
| train_max_steps: 200000 |
| use_fal: false |
| use_fixed_shapes: true |
| use_norm: true |
| var_train_expr: null |
| verbose: 1 |
| version: '0.0' |
|
|
