hyperparams.yaml

# Generated 2023-07-14 from:
# /data2/cloned_repos/speechbrain-clone/recipes/ESC50/interpret/hparams/piq.yaml
# yamllint disable
# #################################
# The recipe for training PIQ on the ESC50 dataset.
#
# Author:
#  * Cem Subakan 2022, 2023
#  * Francesco Paissan 2022, 2023
#  (based on the SpeechBrain UrbanSound8k recipe)
# #################################

# Seed needs to be set at top of yaml, before objects with parameters are made
seed: 1234
__set_seed: !!python/object/apply:torch.manual_seed [1234]

# Set up folders for reading from and writing to
# Dataset must already exist at `audio_data_folder`
data_folder: /data2/ESC-50-master
                          # e.g., /localscratch/UrbanSound8K
audio_data_folder: /data2/ESC-50-master/audio

experiment_name: piq
output_folder: ./results/piq/1234
save_folder: ./results/piq/1234/save
train_log: ./results/piq/1234/train_log.txt

test_only: false
save_interpretations: true
interpret_period: 10

# Tensorboard logs
use_tensorboard: false
tensorboard_logs_folder: ./results/piq/1234/tb_logs/

# Path where data manifest files will be stored
train_annotation: /data2/ESC-50-master/manifest/train.json
valid_annotation: /data2/ESC-50-master/manifest/valid.json
test_annotation: /data2/ESC-50-master/manifest/test.json

# To standardize results, UrbanSound8k has pre-separated samples into
# 10 folds for multi-fold validation
train_fold_nums: [1, 2, 3]
valid_fold_nums: [4]
test_fold_nums: [5]
skip_manifest_creation: false

ckpt_interval_minutes: 15 # save checkpoint every N min

# Training parameters
number_of_epochs: 200
batch_size: 16
lr: 0.0002
sample_rate: 16000
use_vq: true
rec_loss_coef: 1
use_mask_output: true
mask_th: 0.35

device: cuda

# Feature parameters
n_mels: 80

# Number of classes
out_n_neurons: 50

shuffle: true
dataloader_options:
  batch_size: 16
  shuffle: true
  num_workers: 0

epoch_counter: &id001 !new:speechbrain.utils.epoch_loop.EpochCounter

  limit: 200

opt_class: !name:torch.optim.Adam
  lr: 0.0002
  weight_decay: 0.000002

lr_annealing: !new:speechbrain.nnet.schedulers.ReduceLROnPlateau
  factor: 0.5
  patience: 3
  dont_halve_until_epoch: 100

# Logging + checkpoints
train_logger: !new:speechbrain.utils.train_logger.FileTrainLogger
  save_file: ./results/piq/1234/train_log.txt

checkpointer: !new:speechbrain.utils.checkpoints.Checkpointer
  checkpoints_dir: ./results/piq/1234/save
  recoverables:
    psi_model: &id004 !new:speechbrain.lobes.models.PIQ.VectorQuantizedPSI_Audio
      dim: 256
      K: 1024
      shared_keys: 0
      activate_class_partitioning: true
      use_adapter: true
      adapter_reduce_dim: true

    counter: *id001
use_pretrained: true

# embedding_model: !new:custom_models.Conv2dEncoder_v2
embedding_model: &id002 !new:speechbrain.lobes.models.PIQ.Conv2dEncoder_v2
  dim: 256

classifier: &id003 !new:speechbrain.lobes.models.ECAPA_TDNN.Classifier
  input_size: 256
  out_neurons: 50
  lin_blocks: 1


# Interpretation hyperparams
K: 1024

# pre-processing
n_fft: 1024
spec_mag_power: 0.5
hop_length: 11.6099
win_length: 23.2199
compute_stft: &id005 !new:speechbrain.processing.features.STFT
  n_fft: 1024
  hop_length: 11.6099
  win_length: 23.2199
  sample_rate: 16000

compute_fbank: &id006 !new:speechbrain.processing.features.Filterbank
  n_mels: 80
  n_fft: 1024
  sample_rate: 16000

compute_istft: &id007 !new:speechbrain.processing.features.ISTFT
  sample_rate: 16000
  hop_length: 11.6099
  win_length: 23.2199

label_encoder: !new:speechbrain.dataio.encoder.CategoricalEncoder
psi_model: *id004
modules:
  compute_stft: *id005
  compute_fbank: *id006
  compute_istft: *id007
  psi: *id004
  embedding_model: !ref <embedding_model>
  classifier: !ref <classifier>

embedding_model_path: fpaissan/conv2d_us8k/embedding_modelft.ckpt
classifier_model_path: fpaissan/conv2d_us8k/classifier.ckpt
pretrainer: !new:speechbrain.utils.parameter_transfer.Pretrainer
  loadables:
    embedding_model: !ref <embedding_model>
    classifier: !ref <classifier>
    psi: !ref <psi_model>
    label_encoder: !ref <label_encoder>
  paths:
    embedding_model: fpaissan/conv2d_us8k/embedding_modelft.ckpt
    classifier: fpaissan/conv2d_us8k/classifier.ckpt
    psi: /data2/PIQ-ESC50/psi_model.ckpt
    label_encoder: speechbrain/cnn14-esc50/label_encoder.txt