upload train config and vocabulary

train.yaml

@@ -0,0 +1,114 @@
+accum_grad: 8
+cmvn: global_mvn
+cmvn_conf:
+  cmvn_file: /data/models/dolphin-small/feats_stats.npz
+ctc: ctc
+ctc_conf:
+  ctc_blank_id: 0
+dataset: asr
+dataset_conf:
+  batch_conf:
+    batch_size: 16
+    batch_type: static
+  ctc_label: true
+  cycle: 100
+  fbank_conf:
+    dither: 0.1
+    frame_length: 25
+    frame_shift: 10
+    num_mel_bins: 80
+  frontend_conf:
+    n_fft: 512
+    win_length: 400
+    hop_length: 160
+    fs: 16000
+  filter_conf:
+    max_length: 3000
+    min_length: 0
+    token_max_length: 200
+    token_min_length: 1
+  no_time_idx: 3
+  remove_punctuation: true
+  remove_timestamp: true
+  resample_conf:
+    resample_rate: 16000
+  shuffle: true
+  shuffle_conf:
+    shuffle_size: 5120
+  sort: true
+  sort_conf:
+    sort_size: 2048
+  spec_aug: true
+  spec_aug_conf:
+    max_f: 10
+    max_t: 50
+    num_f_mask: 2
+    num_t_mask: 2
+  speed_perturb: true
+  time_apply_prob: 0.0
+decoder: transformer
+decoder_conf:
+  attention_heads: 8
+  dropout_rate: 0.1
+  linear_units: 2048
+  num_blocks: 6
+  positional_dropout_rate: 0.1
+  self_attention_dropout_rate: 0.1
+  src_attention_dropout_rate: 0.1
+  use_sdpa: false
+dtype: fp32
+encoder: e_branchformer
+encoder_conf:
+  activation_type: swish
+  attention_dropout_rate: 0.1
+  attention_heads: 8
+  causal: false
+  cgmlp_conv_kernel: 31
+  cgmlp_linear_units: 2048
+  dropout_rate: 0.1
+  gate_activation: identity
+  input_layer: conv2d
+  linear_units: 2048
+  merge_conv_kernel: 31
+  num_blocks: 6
+  output_size: 512
+  pos_enc_layer_type: rel_pos_v1
+  positional_dropout_rate: 0.1
+  selfattention_layer_type: rel_selfattn
+  use_linear_after_conv: false
+  use_sdpa: false
+grad_clip: 5
+input_dim: 80
+log_interval: 200
+max_epoch: 100
+model: asr_model
+model_conf:
+  ctc_weight: 0.3
+  length_normalized_loss: false
+  lsm_weight: 0.1
+model_dir: exp/dolphin_ebf_small_nonstreaming_v4.3
+optim: adam
+optim_conf:
+  lr: 0.0005
+output_dim: 40002
+save_interval: 1000
+save_states: model_only
+scheduler: warmuplr
+scheduler_conf:
+  warmup_steps: 2048
+stats_dialect: true
+tokenizer: bpe
+tokenizer_conf:
+  bpe_path: /data/models/dolphin-small/bpe.model
+  symbol_table_path: /data/models/dolphin-small/units.txt
+  special_tokens:
+    <asr>: 6
+    <blank>: 0
+    <eos>: 40000
+    <sos>: 39999 
+    <unk>: 1
+train_engine: torch_ddp
+use_amp: false
+vocab_size: 40002
+layer_norm_eps: 1e-12
+support_timestamp: true