Buckets:
| { | |
| "model_type": "autoencoder", | |
| "sample_size": 81920, | |
| "sample_rate": 16000, | |
| "audio_channels": 1, | |
| "model": { | |
| "pretransform":{ | |
| "type": "patched", | |
| "enable_grad": true, | |
| "config": { | |
| "patch_size": 320, | |
| "channels": 1 | |
| } | |
| }, | |
| "encoder": { | |
| "type": "taae", | |
| "config": { | |
| "in_channels": 320, | |
| "channels": 1024, | |
| "c_mults": [1, 1, 1], | |
| "strides": [1, 1, 2], | |
| "latent_dim": 6, | |
| "transformer_depths": [8, 8, 12], | |
| "use_snake": false, | |
| "checkpointing": false, | |
| "conformer": false, | |
| "layer_scale": true, | |
| "sliding_window": [127,128] | |
| } | |
| }, | |
| "decoder": { | |
| "type": "taae", | |
| "config": { | |
| "out_channels": 320, | |
| "channels": 1024, | |
| "c_mults": [1, 1, 1], | |
| "strides": [1, 1, 2], | |
| "latent_dim": 6, | |
| "transformer_depths": [8, 8, 12], | |
| "use_snake": false, | |
| "checkpointing": false, | |
| "conformer": false, | |
| "layer_scale": true, | |
| "sliding_window": [127,128] | |
| } | |
| }, | |
| "bottleneck": { | |
| "type": "dithered_fsq", | |
| "config": { | |
| "dim": 6, | |
| "levels": 17, | |
| "dither_inference": false, | |
| "num_codebooks": 1, | |
| "noise_dropout": 0.5 | |
| } | |
| }, | |
| "latent_dim": 6, | |
| "downsampling_ratio": 640, | |
| "io_channels": 1 | |
| }, | |
| "training": { | |
| "learning_rate": 1e-4, | |
| "warmup_steps": 0, | |
| "use_ema": true, | |
| "clip_grad_norm": 0.4, | |
| "optimizer_configs": { | |
| "autoencoder": { | |
| "optimizer": { | |
| "type": "AdamW", | |
| "config": { | |
| "betas": [0.9, 0.95], | |
| "lr": 1e-4, | |
| "weight_decay": 1e-2 | |
| } | |
| }, | |
| "scheduler": { | |
| "type": "InverseLR", | |
| "config": { | |
| "inv_gamma": 50000, | |
| "power": 0.5, | |
| "warmup": 0.999 | |
| } | |
| } | |
| }, | |
| "discriminator": { | |
| "optimizer": { | |
| "type": "AdamW", | |
| "config": { | |
| "betas": [0.9, 0.999], | |
| "lr": 1e-4, | |
| "weight_decay": 1e-5 | |
| } | |
| }, | |
| "scheduler": { | |
| "type": "InverseLR", | |
| "config": { | |
| "inv_gamma": 50000, | |
| "power": 0.5, | |
| "warmup": 0.999 | |
| } | |
| } | |
| } | |
| }, | |
| "loss_configs": { | |
| "discriminator": { | |
| "type": "encodec", | |
| "config": { | |
| "filters": 256, | |
| "n_ffts": [206, 334, 542, 876, 1418, 2296], | |
| "hop_lengths": [103, 167, 271, 438, 709, 1148], | |
| "win_lengths": [206, 334, 542, 876, 1418, 2296], | |
| "normalize_losses": true, | |
| "spec_scale_pow": 0.5 | |
| }, | |
| "weights": { | |
| "adversarial": 0.0, | |
| "feature_matching": 1.0 | |
| } | |
| }, | |
| "spectral": { | |
| "type": "mrstft", | |
| "decay": 0.9999, | |
| "config": { | |
| "fft_sizes": [2048], | |
| "hop_sizes": [512], | |
| "win_lengths": [2048] | |
| }, | |
| "weights": { | |
| "mrstft": 1.0 | |
| } | |
| }, | |
| "time": { | |
| "type": "l1/l2", | |
| "decay": 0.9999, | |
| "weights": { | |
| "l1": 1.0, | |
| "l2": 0.0 | |
| } | |
| } | |
| }, | |
| "eval_loss_configs": { | |
| "pesq": {}, | |
| "stft": {}, | |
| "mel": {}, | |
| "sisdr": {}, | |
| "per": {} | |
| }, | |
| "demo": { | |
| "demo_every": 2000 | |
| } | |
| } | |
| } | |
Xet Storage Details
- Size:
- 4.49 kB
- Xet hash:
- 9fa61a67eb1940502a5f0a0bf82b04a165f770a67b32dd1945458d2d1e33208a
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.