Introduce code from Rust monorepo. This is a mess

2026-10-23 08:26:06 +00:00 · 2022-02-28 05:43:54 -05:00
parent 921fb49f9a
commit ba326bfcfd
1058 changed files with 192192 additions and 0 deletions
@@ -0,0 +1,201 @@
+# It contains the default values for training a Conformer ASR model medium size with RNNT loss and BPE-based vocabulary.
+# Default learning parameters in this config are set for effective batch size of 2K.
+# To train it with smaller batch sizes, you may need to re-tune the learning parameters or use higher accumulate_grad_batches.
+
+# You may find more info about Conformer-CTC here: https://docs.nvidia.com/deeplearning/nemo/user-guide/docs/en/stable/asr/models.html#conformer-ctc
+# Pre-trained models of Conformer-CTC can be found here: https://docs.nvidia.com/deeplearning/nemo/user-guide/docs/en/stable/asr/results.html
+
+name: "Conformer-RNNT-BPE"
+
+model:
+  sample_rate: &sample_rate 16000
+  compute_eval_loss: false # disables loss evaluation for validation and test data
+  log_prediction: true # enables logging sample predictions in the output during training
+  ctc_reduction: 'mean_batch'
+
+  model_defaults:
+    se: true
+    se_context_size: -1
+    # encoder / decoder / joint values
+    enc_hidden: ${model.encoder.d_model}
+    pred_hidden: 640
+    joint_hidden: 640
+
+  train_ds:
+    manifest_filepath: ???
+    sample_rate: ${model.sample_rate}
+    batch_size: 16 # you may increase batch_size if your memory allows
+    shuffle: true
+    num_workers: 8
+    pin_memory: false
+    use_start_end_token: true
+    trim_silence: false
+    max_duration: 16.7 # it is set for LibriSpeech, you may need to update it for your dataset
+    min_duration: 0.1
+
+  validation_ds:
+    manifest_filepath: ???
+    sample_rate: ${model.sample_rate}
+    batch_size: 16
+    shuffle: false
+    num_workers: 8
+    pin_memory: false
+    use_start_end_token: true
+
+  test_ds:
+    manifest_filepath: null
+    sample_rate: ${model.sample_rate}
+    batch_size: 16
+    shuffle: false
+    num_workers: 8
+    pin_memory: false
+    use_start_end_token: true
+
+  # you may find more detail on how to train a tokenizer at: /scripts/tokenizers/process_asr_text_tokenizer.py
+  tokenizer:
+    dir: ???  # path to directory which contains either tokenizer.model (bpe) or vocab.txt (for wpe)
+    type: wpe  # Can be either bpe (SentencePiece tokenizer) or wpe (WordPiece tokenizer)
+
+  preprocessor:
+    _target_: nemo.collections.asr.modules.AudioToMelSpectrogramPreprocessor
+    sample_rate: *sample_rate
+    normalize: "per_feature"
+    window_size: 0.025
+    window_stride: 0.01
+    window: "hann"
+    features: 80
+    n_fft: 512
+    frame_splicing: 1
+    dither: 0.00001
+    pad_to: 0
+
+  spec_augment:
+    _target_: nemo.collections.asr.modules.SpectrogramAugmentation
+    freq_masks: 2 # set to zero to disable it
+    time_masks: 10 # set to zero to disable it
+    freq_width: 27
+    time_width: 0.05
+
+  encoder:
+    _target_: nemo.collections.asr.modules.ConformerEncoder
+    feat_in: ${model.preprocessor.features}
+    feat_out: -1 # you may set it if you need different output size other than the default d_model
+    n_layers: 16
+    d_model: 256
+
+    # Sub-sampling params
+    subsampling: striding # vggnet or striding
+    subsampling_factor: 4 # must be power of 2
+    subsampling_conv_channels: -1 # set to -1 to make it equal to the d_model
+
+    # Feed forward module's params
+    ff_expansion_factor: 4
+
+    # Multi-headed Attention Module's params
+    self_attention_model: rel_pos # rel_pos or abs_pos
+    n_heads: 4
+    xscaling: true # scales up the inputs by sqrt(d_model)
+    untie_biases: true # unties the biases of the TransformerXL layers
+    pos_emb_max_len: 5000
+
+    # Convolution module's params
+    conv_kernel_size: 31
+
+    ### regularization
+    dropout: 0.1 # The dropout used in most of the Conformer Modules
+    dropout_emb: 0.0 # The dropout used for embeddings
+    dropout_att: 0.1 # The dropout for multi-headed attention modules
+
+  decoder:
+    _target_: nemo.collections.asr.modules.RNNTDecoder
+    normalization_mode: null
+    random_state_sampling: false
+    blank_as_pad: true
+
+    prednet:
+      pred_hidden: ${model.model_defaults.pred_hidden}
+      pred_rnn_layers: 1
+      t_max: null
+      dropout: 0.1
+
+  joint:
+    _target_: nemo.collections.asr.modules.RNNTJoint
+    log_softmax: null  # 'null' would set it automatically according to CPU/GPU device
+
+    # fused mode
+    experimental_fuse_loss_wer: true
+    fused_batch_size: 16
+
+    jointnet:
+      joint_hidden: ${model.model_defaults.joint_hidden}
+      activation: "relu"
+      dropout: 0.1
+
+  decoding:
+    strategy: "greedy_batch"
+
+    # greedy strategy config
+    greedy:
+      max_symbols: 30
+
+    # beam strategy config
+    beam:
+      beam_size: 2
+      score_norm: true
+      tsd_max_sym_exp: 50  # for Time Synchronous Decoding
+      alsd_max_target_len: 2.0  # for Alignment-Length Synchronous Decoding
+
+  optim:
+    name: adam
+    lr: 5.0
+    # optimizer arguments
+    betas: [0.9, 0.98]
+    weight_decay: 1e-6
+
+    # scheduler setup
+    sched:
+      name: NoamAnnealing
+      d_model: ${model.encoder.d_model}
+      # scheduler config override
+      warmup_steps: 10000
+      warmup_ratio: null
+      min_lr: 1e-6
+
+trainer:
+  gpus: -1 # number of GPUs, -1 would use all available GPUs
+  num_nodes: 1
+  max_epochs: 1000
+  max_steps: null # 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: ddp
+  accumulate_grad_batches: 1
+  gradient_clip_val: 0.0
+  amp_level: O0 # O1/O2 for mixed precision
+  precision: 32 # Should be set to 16 for O1 and O2 to enable the AMP.
+  log_every_n_steps: 10  # Interval of logging.
+  progress_bar_refresh_rate: 10
+  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
+  checkpoint_callback: false  # Provided by exp_manager
+  logger: false  # Provided by exp_manager
+
+
+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_wer"
+    mode: "min"
+    save_top_k: 3
+  resume_if_exists: false
+  resume_ignore_no_checkpoint: false
+
+  create_wandb_logger: false
+  wandb_logger_kwargs:
+    name: null
+    project: null