File size: 17,376 Bytes
7934b29 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 |
# This config contains the default values for self-supervised pre-training of ContextNet encoder.
# In contrast to original ContextNet, the same number of filters is used throughout the model.
# Default learning parameters in this config are set for effective batch size of 1K. To train it with smaller effective
# batch sizes, you may need to re-tune the learning parameters or use higher accumulate_grad_batches.
# Here are the recommended configs for different variants of ContextNet, other parameters are the same as in this config file.
#
# +-------------+---------+------------+
# | Model | filters | time_masks |
# +=============+=========+============+
# | Small (14M)| 256 | 2 |
# +-------------+---------+------------+
# | Medium (40M)| 512 | 5 |
# +-------------+---------+------------+
# | Large (145M)| 1024 | 10 |
# +-------------------------------------
name: &name "ContextNet-8x-Stride-SSL"
model:
sample_rate: &sample_rate 16000
train_ds:
manifest_filepath: ???
sample_rate: ${model.sample_rate}
batch_size: 16 # Can be increased if memory allows or when using smaller model
trim_silence: false
max_duration: 16.7
min_duration: 8.0
shuffle: true
use_start_end_token: false
num_workers: 16
pin_memory: true
# tarred datasets
is_tarred: false
tarred_audio_filepaths: null
tarred_shard_strategy: "scatter"
shuffle_n: 2048
# bucketing params
bucketing_strategy: "synced_randomized"
bucketing_batch_size: null
validation_ds:
manifest_filepath: ???
sample_rate: ${model.sample_rate}
batch_size: 8
shuffle: false
use_start_end_token: false
num_workers: 16
pin_memory: true
min_duration: 8.0
model_defaults:
filters: 1024
repeat: 5
dropout: 0.1
separable: true
se: true
se_context_size: -1
kernel_size_factor: 1.0
enc_hidden: 640
decoder_out_channels: 128
preprocessor:
_target_: nemo.collections.asr.modules.AudioToMelSpectrogramPreprocessor
sample_rate: ${model.sample_rate}
normalize: "per_feature"
window_size: 0.025
window_stride: 0.01
window: "hann"
features: &n_mels 80
n_fft: 512
frame_splicing: 1
dither: 0.00001
pad_to: 16
stft_conv: false
spec_augment:
_target_: nemo.collections.asr.modules.MaskedPatchAugmentation
freq_masks: 3
freq_width: 20
patch_size: 48
mask_patches: 0.5
encoder:
_target_: nemo.collections.asr.modules.ConvASREncoder
feat_in: *n_mels
activation: swish
conv_mask: true
init_mode: "tds_uniform"
jasper:
- filters: ${model.model_defaults.filters}
repeat: 1
kernel: [5]
stride: [1]
dilation: [1]
dropout: 0.0
residual: false
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [2]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
stride_last: true
residual_mode: "stride_add"
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [2] # *stride
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
stride_last: true
residual_mode: "stride_add"
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [2] # stride
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
stride_last: true
residual_mode: "stride_add"
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.filters}
repeat: ${model.model_defaults.repeat}
kernel: [5]
stride: [1]
dilation: [1]
dropout: ${model.model_defaults.dropout}
residual: true
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
- filters: ${model.model_defaults.enc_hidden}
repeat: 1
kernel: [5]
stride: [1]
dilation: [1]
dropout: 0.0
residual: false
separable: ${model.model_defaults.separable}
se: ${model.model_defaults.se}
se_context_size: ${model.model_defaults.se_context_size}
kernel_size_factor: ${model.model_defaults.kernel_size_factor}
loss_list:
contrastive:
decoder:
_target_: nemo.collections.asr.modules.ConvASRDecoderReconstruction
feat_in: ${model.model_defaults.enc_hidden}
feat_hidden: 128
# features in hidden layer of decoder
feat_out: ${model.model_defaults.decoder_out_channels}
stride_layers: 1
# if loss.combine_time_steps is different than the encoder stride,
# then a corresponding amount of stride_layers needs to
# be added to the decoder (here stride is 8 and combine_time_steps is 4)
non_stride_layers: 0
stride_transpose: true
apply_softmax: false
loss:
_target_: nemo.collections.asr.losses.ContrastiveLoss
in_dim: ${model.preprocessor.features}
proj_dim: ${model.model_defaults.decoder_out_channels}
combine_time_steps: 4 #how many spectrogram time steps are used for one target/representation for contrastive task
quantized_targets: true #should quantizer or linear layer be used
# (quantizer is required to extract pseudo-labels for other losses)
codebook_size: 300 # number of vectors in the quantization codebook per group
num_groups: 2 # number of groups in the quantizer codebook
num_negatives: 100 # number of sampled negatives for each target
sample_from_same_utterance_only: true #should negatives be sampled only from the same utterance
sample_from_non_masked: false #should negatives be sampled from non-masked steps
mlm:
decoder:
_target_: nemo.collections.asr.modules.ConvASRDecoderReconstruction
feat_in: ${model.model_defaults.enc_hidden}
feat_hidden: 128
# features in hidden layer of decoder
feat_out: 90000
# this should be equal to codebook_size^groups in the contrastive loss to match the targets
stride_layers: 1
stride_transpose: true
activation: "identity"
apply_softmax: true
loss:
_target_: nemo.collections.asr.losses.MLMLoss
combine_time_steps: 4
targets_from_loss: "contrastive"
loss_alpha: 1000.
optim:
name: adamw
lr: 5.0
# optimizer arguments
betas: [0.9, 0.98]
weight_decay: 1e-3
# scheduler setup
sched:
name: NoamAnnealing
d_model: ${model.model_defaults.enc_hidden}
# scheduler config override
warmup_steps: 25000
warmup_ratio: null
min_lr: 1e-6
trainer:
devices: -1 # number of GPUs, -1 would use all available GPUs
num_nodes: 1
max_epochs: 1000
max_steps: -1 # computed at runtime if not set
val_check_interval: 1.0 # Set to 0.25 to check 4 times per epoch, or an int for number of iterations
accelerator: auto
strategy: ddp
accumulate_grad_batches: 1
gradient_clip_val: 1.0
precision: 32 # Should be set to 16 for O1 and O2 to enable the AMP.
log_every_n_steps: 10 # Interval of logging.
enable_progress_bar: True
resume_from_checkpoint: null # The path to a checkpoint file to continue the training, restores the whole state including the epoch, step, LR schedulers, apex, etc.
num_sanity_val_steps: 0 # number of steps to perform validation steps for sanity check the validation process before starting the training, setting to 0 disables it
check_val_every_n_epoch: 1 # number of evaluations on validation every n epochs
sync_batchnorm: true
enable_checkpointing: False # Provided by exp_manager
logger: false # Provided by exp_manager
benchmark: false # needs to be false for models with variable-length speech input as it slows down training
exp_manager:
exp_dir: null
name: ${name}
create_tensorboard_logger: true
create_checkpoint_callback: true
checkpoint_callback_params:
# in case of multiple validation sets, first one is used
monitor: "val_loss"
mode: "min"
save_top_k: 5
# you need to set these two to True to continue the training
resume_if_exists: false
resume_ignore_no_checkpoint: false
# You may use this section to create a W&B logger
create_wandb_logger: false
wandb_logger_kwargs:
name: null
project: null
|