|
accelerator: gpu |
|
accumulate_grad_batches: null |
|
amp_backend: native |
|
amp_level: null |
|
auto_lr_find: false |
|
auto_scale_batch_size: false |
|
auto_select_gpus: false |
|
batch_size: 4 |
|
benchmark: null |
|
betas: !!python/tuple |
|
- 0.8 |
|
- 0.99 |
|
c_kl: 1.0 |
|
c_mel: 45 |
|
channels: 1 |
|
check_val_every_n_epoch: 1 |
|
checkpoint_epochs: 1 |
|
dataset: |
|
- !!python/object/apply:pathlib.PosixPath |
|
- / |
|
- kaggle |
|
- working |
|
- piper |
|
- Test-high-cleaned-210-samples |
|
- dataset.jsonl |
|
dataset_dir: &id001 !!python/object/apply:pathlib.PosixPath |
|
- / |
|
- kaggle |
|
- working |
|
- piper |
|
- Test-high-cleaned-210-samples |
|
default_root_dir: *id001 |
|
detect_anomaly: false |
|
deterministic: null |
|
devices: '1' |
|
enable_checkpointing: true |
|
enable_model_summary: true |
|
enable_progress_bar: true |
|
eps: 1.0e-09 |
|
fast_dev_run: false |
|
filter_channels: 768 |
|
filter_length: 1024 |
|
gin_channels: 0 |
|
gpus: null |
|
grad_clip: null |
|
gradient_clip_algorithm: null |
|
gradient_clip_val: null |
|
hidden_channels: 192 |
|
hop_length: 256 |
|
init_lr_ratio: 1.0 |
|
inter_channels: 192 |
|
ipus: null |
|
kernel_size: 3 |
|
learning_rate: 0.0002 |
|
limit_predict_batches: null |
|
limit_test_batches: null |
|
limit_train_batches: null |
|
limit_val_batches: null |
|
log_every_n_steps: 10 |
|
logger: true |
|
lr_decay: 0.999875 |
|
max_epochs: 200 |
|
max_phoneme_ids: null |
|
max_steps: -1 |
|
max_time: null |
|
mel_channels: 80 |
|
mel_fmax: null |
|
mel_fmin: 0.0 |
|
min_epochs: null |
|
min_steps: null |
|
move_metrics_to_cpu: false |
|
multiple_trainloader_mode: max_size_cycle |
|
n_heads: 2 |
|
n_layers: 6 |
|
n_layers_q: 3 |
|
num_ckpt: 0 |
|
num_nodes: 1 |
|
num_processes: null |
|
num_sanity_val_steps: 2 |
|
num_speakers: 2 |
|
num_symbols: 256 |
|
num_test_examples: 0 |
|
num_workers: 1 |
|
overfit_batches: 0.0 |
|
p_dropout: 0.1 |
|
plugins: null |
|
precision: 32 |
|
profiler: null |
|
quality: high |
|
reload_dataloaders_every_n_epochs: 0 |
|
replace_sampler_ddp: true |
|
resblock: '1' |
|
resblock_dilation_sizes: !!python/tuple |
|
- &id002 !!python/tuple |
|
- 1 |
|
- 3 |
|
- 5 |
|
- *id002 |
|
- *id002 |
|
resblock_kernel_sizes: !!python/tuple |
|
- 3 |
|
- 7 |
|
- 11 |
|
resume_from_checkpoint: null |
|
resume_from_single_speaker_checkpoint: /content/pretrained.ckpt |
|
sample_bytes: 2 |
|
sample_rate: 22050 |
|
save_last: true |
|
seed: 1234 |
|
segment_size: 8192 |
|
strategy: null |
|
sync_batchnorm: false |
|
tpu_cores: null |
|
track_grad_norm: -1 |
|
upsample_initial_channel: 512 |
|
upsample_kernel_sizes: !!python/tuple |
|
- 16 |
|
- 16 |
|
- 4 |
|
- 4 |
|
upsample_rates: !!python/tuple |
|
- 8 |
|
- 8 |
|
- 2 |
|
- 2 |
|
use_sdp: true |
|
use_spectral_norm: false |
|
val_check_interval: null |
|
validation_split: 0.0 |
|
warmup_epochs: 0 |
|
weights_save_path: null |
|
win_length: 1024 |
|
|
