Update Model

exp/asr_stats_raw_en_word/logdir/q/stats.log

@@ -0,0 +1 @@
+Submitted batch job 5730432

exp/asr_stats_raw_en_word/logdir/q/stats.sh

@@ -0,0 +1,29 @@
+#!/bin/bash
+cd /ocean/projects/cis210027p/ganesank/karthik_new/espnet/egs2/sinhala/asr1
+. ./path.sh
+( echo '#' Running on `hostname`
+  echo '#' Started at `date`
+  set | grep SLURM | while read line; do echo "# $line"; done
+  echo -n '# '; cat <<EOF
+python3 -m espnet2.bin.asr_train --collect_stats true --use_preprocessor true --bpemodel none --token_type word --token_list data/en_token_list/word/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --train_data_path_and_name_and_type dump/raw/train/wav.scp,speech,sound --train_data_path_and_name_and_type dump/raw/train/text,text,text --valid_data_path_and_name_and_type dump/raw/valid/wav.scp,speech,sound --valid_data_path_and_name_and_type dump/raw/valid/text,text,text --train_shape_file exp/asr_stats_raw_en_word/logdir/train.${SLURM_ARRAY_TASK_ID}.scp --valid_shape_file exp/asr_stats_raw_en_word/logdir/valid.${SLURM_ARRAY_TASK_ID}.scp --output_dir exp/asr_stats_raw_en_word/logdir/stats.${SLURM_ARRAY_TASK_ID} --config conf/train_asr.yaml --frontend_conf fs=16k 
+EOF
+) >exp/asr_stats_raw_en_word/logdir/stats.$SLURM_ARRAY_TASK_ID.log
+if [ "$CUDA_VISIBLE_DEVICES" == "NoDevFiles" ]; then
+  ( echo CUDA_VISIBLE_DEVICES set to NoDevFiles, unsetting it... 
+  )>>exp/asr_stats_raw_en_word/logdir/stats.$SLURM_ARRAY_TASK_ID.log
+  unset CUDA_VISIBLE_DEVICES
+fi
+time1=`date +"%s"`
+ ( python3 -m espnet2.bin.asr_train --collect_stats true --use_preprocessor true --bpemodel none --token_type word --token_list data/en_token_list/word/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --train_data_path_and_name_and_type dump/raw/train/wav.scp,speech,sound --train_data_path_and_name_and_type dump/raw/train/text,text,text --valid_data_path_and_name_and_type dump/raw/valid/wav.scp,speech,sound --valid_data_path_and_name_and_type dump/raw/valid/text,text,text --train_shape_file exp/asr_stats_raw_en_word/logdir/train.${SLURM_ARRAY_TASK_ID}.scp --valid_shape_file exp/asr_stats_raw_en_word/logdir/valid.${SLURM_ARRAY_TASK_ID}.scp --output_dir exp/asr_stats_raw_en_word/logdir/stats.${SLURM_ARRAY_TASK_ID} --config conf/train_asr.yaml --frontend_conf fs=16k  ) &>>exp/asr_stats_raw_en_word/logdir/stats.$SLURM_ARRAY_TASK_ID.log
+ret=$?
+sync || true
+time2=`date +"%s"`
+echo '#' Accounting: begin_time=$time1 >>exp/asr_stats_raw_en_word/logdir/stats.$SLURM_ARRAY_TASK_ID.log
+echo '#' Accounting: end_time=$time2 >>exp/asr_stats_raw_en_word/logdir/stats.$SLURM_ARRAY_TASK_ID.log
+echo '#' Accounting: time=$(($time2-$time1)) threads=1 >>exp/asr_stats_raw_en_word/logdir/stats.$SLURM_ARRAY_TASK_ID.log
+echo '#' Finished at `date` with status $ret >>exp/asr_stats_raw_en_word/logdir/stats.$SLURM_ARRAY_TASK_ID.log
+[ $ret -eq 137 ] && exit 100;
+touch exp/asr_stats_raw_en_word/logdir/q/done.1797611.$SLURM_ARRAY_TASK_ID
+exit $[$ret ? 1 : 0]
+## submitted with:
+# sbatch --export=PATH  --time 10:00:00 -p RM-shared  --open-mode=append -e exp/asr_stats_raw_en_word/logdir/q/stats.log -o exp/asr_stats_raw_en_word/logdir/q/stats.log --array 1-32 /ocean/projects/cis210027p/ganesank/karthik_new/espnet/egs2/sinhala/asr1/exp/asr_stats_raw_en_word/logdir/q/stats.sh >>exp/asr_stats_raw_en_word/logdir/q/stats.log 2>&1

exp/asr_stats_raw_en_word/logdir/stats.1.log

@@ -0,0 +1,495 @@
+# Running on r041.ib.bridges2.psc.edu
+# Started at Tue Dec 21 22:23:13 EST 2021
+# SLURMD_NODENAME=r041
+# SLURM_ARRAY_JOB_ID=5730432
+# SLURM_ARRAY_TASK_COUNT=32
+# SLURM_ARRAY_TASK_ID=1
+# SLURM_ARRAY_TASK_MAX=32
+# SLURM_ARRAY_TASK_MIN=1
+# SLURM_ARRAY_TASK_STEP=1
+# SLURM_CLUSTER_NAME=bridges2
+# SLURM_CONF=/var/spool/slurm/d/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=1
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GTIDS=0
+# SLURM_JOBID=5730448
+# SLURM_JOB_ACCOUNT=cis210027p
+# SLURM_JOB_CPUS_PER_NODE=1
+# SLURM_JOB_GID=24886
+# SLURM_JOB_ID=5730448
+# SLURM_JOB_NAME=stats.sh
+# SLURM_JOB_NODELIST=r041
+# SLURM_JOB_NUM_NODES=1
+# SLURM_JOB_PARTITION=RM-shared
+# SLURM_JOB_QOS=rm
+# SLURM_JOB_UID=82326
+# SLURM_JOB_USER=ganesank
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_CPU=2000
+# SLURM_NNODES=1
+# SLURM_NODEID=0
+# SLURM_NODELIST=r041
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/ocean/projects/cis210027p/ganesank/karthik_new/espnet/egs2/sinhala/asr1
+# SLURM_SUBMIT_HOST=br012.ib.bridges2.psc.edu
+# SLURM_TASKS_PER_NODE=1
+# SLURM_TASK_PID=98449
+# SLURM_TOPOLOGY_ADDR=r041
+# SLURM_TOPOLOGY_ADDR_PATTERN=node
+# SLURM_WORKING_CLUSTER=bridges2:br003:6814:9216:109
+# python3 -m espnet2.bin.asr_train --collect_stats true --use_preprocessor true --bpemodel none --token_type word --token_list data/en_token_list/word/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --train_data_path_and_name_and_type dump/raw/train/wav.scp,speech,sound --train_data_path_and_name_and_type dump/raw/train/text,text,text --valid_data_path_and_name_and_type dump/raw/valid/wav.scp,speech,sound --valid_data_path_and_name_and_type dump/raw/valid/text,text,text --train_shape_file exp/asr_stats_raw_en_word/logdir/train.1.scp --valid_shape_file exp/asr_stats_raw_en_word/logdir/valid.1.scp --output_dir exp/asr_stats_raw_en_word/logdir/stats.1 --config conf/train_asr.yaml --frontend_conf fs=16k 
+/ocean/projects/cis210027p/ganesank/karthik_new/espnet/tools/venv/bin/python3 /ocean/projects/cis210027p/ganesank/karthik_new/espnet/espnet2/bin/asr_train.py --collect_stats true --use_preprocessor true --bpemodel none --token_type word --token_list data/en_token_list/word/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --train_data_path_and_name_and_type dump/raw/train/wav.scp,speech,sound --train_data_path_and_name_and_type dump/raw/train/text,text,text --valid_data_path_and_name_and_type dump/raw/valid/wav.scp,speech,sound --valid_data_path_and_name_and_type dump/raw/valid/text,text,text --train_shape_file exp/asr_stats_raw_en_word/logdir/train.1.scp --valid_shape_file exp/asr_stats_raw_en_word/logdir/valid.1.scp --output_dir exp/asr_stats_raw_en_word/logdir/stats.1 --config conf/train_asr.yaml --frontend_conf fs=16k
+[r041] 2021-12-21 22:24:17,989 (asr:382) INFO: Vocabulary size: 40
+[r041] 2021-12-21 22:24:18,946 (abs_task:1132) INFO: pytorch.version=1.8.1+cu102, cuda.available=False, cudnn.version=7605, cudnn.benchmark=False, cudnn.deterministic=True
+[r041] 2021-12-21 22:24:18,952 (abs_task:1133) INFO: Model structure:
+ESPnetASRModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=512, hop_length=128, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (time_warp): TimeWarp(window=5, mode=bicubic)
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 30], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxis(mask_width_range=[0, 40], num_mask=2, axis=time)
+  )
+  (normalize): UtteranceMVN(norm_means=True, norm_vars=False)
+  (encoder): TransformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 256, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=4864, out_features=256, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (9): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (10): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (11): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(40, 256)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=256, out_features=40, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=256, out_features=40, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetASRModel
+    Total Number of model parameters: 27.12 M
+    Number of trainable parameters: 27.12 M (100.0%)
+    Size: 108.49 MB
+    Type: torch.float32
+[r041] 2021-12-21 22:24:18,952 (abs_task:1136) INFO: Optimizer:
+Adam (
+Parameter Group 0
+    amsgrad: False
+    betas: (0.9, 0.999)
+    eps: 1e-08
+    initial_lr: 0.0002
+    lr: 8e-09
+    weight_decay: 0
+)
+[r041] 2021-12-21 22:24:18,952 (abs_task:1137) INFO: Scheduler: WarmupLR(warmup_steps=25000)
+[r041] 2021-12-21 22:24:18,954 (abs_task:1146) INFO: Saving the configuration in exp/asr_stats_raw_en_word/logdir/stats.1/config.yaml
+[r041] 2021-12-21 22:24:18,966 (abs_task:1157) INFO: Namespace(config='conf/train_asr.yaml', print_config=False, log_level='INFO', dry_run=False, iterator_type='sequence', output_dir='exp/asr_stats_raw_en_word/logdir/stats.1', ngpu=0, seed=0, num_workers=1, num_att_plot=3, dist_backend='nccl', dist_init_method='env://', dist_world_size=None, dist_rank=None, local_rank=None, dist_master_addr=None, dist_master_port=None, dist_launcher=None, multiprocessing_distributed=False, unused_parameters=False, sharded_ddp=False, cudnn_enabled=True, cudnn_benchmark=False, cudnn_deterministic=True, collect_stats=True, write_collected_feats=False, max_epoch=50, patience=None, val_scheduler_criterion=('valid', 'loss'), early_stopping_criterion=('valid', 'loss', 'min'), best_model_criterion=[('train', 'loss', 'min'), ('valid', 'loss', 'min'), ('train', 'acc', 'max'), ('valid', 'acc', 'max')], keep_nbest_models=5, grad_clip=5.0, grad_clip_type=2.0, grad_noise=False, accum_grad=1, no_forward_run=False, resume=False, train_dtype='float32', use_amp=False, log_interval=None, use_tensorboard=True, use_wandb=False, wandb_project=None, wandb_id=None, wandb_entity=None, wandb_name=None, wandb_model_log_interval=-1, detect_anomaly=False, pretrain_path=None, init_param=[], ignore_init_mismatch=False, freeze_param=[], num_iters_per_epoch=None, batch_size=20, valid_batch_size=None, batch_bins=1000000, valid_batch_bins=None, train_shape_file=['exp/asr_stats_raw_en_word/logdir/train.1.scp'], valid_shape_file=['exp/asr_stats_raw_en_word/logdir/valid.1.scp'], batch_type='folded', valid_batch_type=None, fold_length=[], sort_in_batch='descending', sort_batch='descending', multiple_iterator=False, chunk_length=500, chunk_shift_ratio=0.5, num_cache_chunks=1024, train_data_path_and_name_and_type=[('dump/raw/train/wav.scp', 'speech', 'sound'), ('dump/raw/train/text', 'text', 'text')], valid_data_path_and_name_and_type=[('dump/raw/valid/wav.scp', 'speech', 'sound'), ('dump/raw/valid/text', 'text', 'text')], allow_variable_data_keys=False, max_cache_size=0.0, max_cache_fd=32, valid_max_cache_size=None, optim='adam', optim_conf={'lr': 0.0002}, scheduler='warmuplr', scheduler_conf={'warmup_steps': 25000}, token_list=['<blank>', '<unk>', '්', 'න', 'ම', 'ක', 'ල', 'ි', 'ු', 'ග', 'ේ', 'ර', 'ත', 'ද', 'ව', 'ට', 'ඕ', 'ී', 'ප', 'ය', 'ෙ', 'ස', 'ණ', 'ා', 'ැ', 'RequestAcc.balance', 'Moneywithdraw', 'Moneydeposit', 'Moneytransfer', 'Billpayments', 'බ', 'ඉ', 'ශ', 'ෂ', 'ඩ', 'Creditcardpayments', 'එ', '\u200d', 'හ', '<sos/eos>'], init=None, input_size=None, ctc_conf={'dropout_rate': 0.0, 'ctc_type': 'builtin', 'reduce': True, 'ignore_nan_grad': True}, model_conf={'ctc_weight': 0.5, 'ignore_id': -1, 'lsm_weight': 0.0, 'length_normalized_loss': False, 'report_cer': True, 'report_wer': True, 'sym_space': '<space>', 'sym_blank': '<blank>', 'extract_feats_in_collect_stats': True}, use_preprocessor=True, token_type='word', bpemodel=None, non_linguistic_symbols=None, cleaner=None, g2p=None, speech_volume_normalize=None, rir_scp=None, rir_apply_prob=1.0, noise_scp=None, noise_apply_prob=1.0, noise_db_range='13_15', frontend='default', frontend_conf={'fs': '16k'}, specaug='specaug', specaug_conf={'apply_time_warp': True, 'time_warp_window': 5, 'time_warp_mode': 'bicubic', 'apply_freq_mask': True, 'freq_mask_width_range': [0, 30], 'num_freq_mask': 2, 'apply_time_mask': True, 'time_mask_width_range': [0, 40], 'num_time_mask': 2}, normalize='utterance_mvn', normalize_conf={}, preencoder=None, preencoder_conf={}, encoder='transformer', encoder_conf={'output_size': 256, 'attention_heads': 4, 'linear_units': 2048, 'num_blocks': 12, 'dropout_rate': 0.1, 'positional_dropout_rate': 0.1, 'attention_dropout_rate': 0.0, 'input_layer': 'conv2d', 'normalize_before': True}, postencoder=None, postencoder_conf={}, decoder='transformer', decoder_conf={'attention_heads': 4, 'linear_units': 2048, 'num_blocks': 6, 'dropout_rate': 0.1, 'positional_dropout_rate': 0.1, 'self_attention_dropout_rate': 0.0, 'src_attention_dropout_rate': 0.0}, required=['output_dir', 'token_list'], version='0.10.3a3', distributed=False)
+# Accounting: begin_time=1640143393
+# Accounting: end_time=1640143471
+# Accounting: time=78 threads=1
+# Finished at Tue Dec 21 22:24:31 EST 2021 with status 0

exp/asr_stats_raw_en_word/logdir/stats.1/config.yaml

@@ -0,0 +1,231 @@
+config: conf/train_asr.yaml
+print_config: false
+log_level: INFO
+dry_run: false
+iterator_type: sequence
+output_dir: exp/asr_stats_raw_en_word/logdir/stats.1
+ngpu: 0
+seed: 0
+num_workers: 1
+num_att_plot: 3
+dist_backend: nccl
+dist_init_method: env://
+dist_world_size: null
+dist_rank: null
+local_rank: null
+dist_master_addr: null
+dist_master_port: null
+dist_launcher: null
+multiprocessing_distributed: false
+unused_parameters: false
+sharded_ddp: false
+cudnn_enabled: true
+cudnn_benchmark: false
+cudnn_deterministic: true
+collect_stats: true
+write_collected_feats: false
+max_epoch: 50
+patience: null
+val_scheduler_criterion:
+- valid
+- loss
+early_stopping_criterion:
+- valid
+- loss
+- min
+best_model_criterion:
+-   - train
+    - loss
+    - min
+-   - valid
+    - loss
+    - min
+-   - train
+    - acc
+    - max
+-   - valid
+    - acc
+    - max
+keep_nbest_models: 5
+grad_clip: 5.0
+grad_clip_type: 2.0
+grad_noise: false
+accum_grad: 1
+no_forward_run: false
+resume: false
+train_dtype: float32
+use_amp: false
+log_interval: null
+use_tensorboard: true
+use_wandb: false
+wandb_project: null
+wandb_id: null
+wandb_entity: null
+wandb_name: null
+wandb_model_log_interval: -1
+detect_anomaly: false
+pretrain_path: null
+init_param: []
+ignore_init_mismatch: false
+freeze_param: []
+num_iters_per_epoch: null
+batch_size: 20
+valid_batch_size: null
+batch_bins: 1000000
+valid_batch_bins: null
+train_shape_file:
+- exp/asr_stats_raw_en_word/logdir/train.1.scp
+valid_shape_file:
+- exp/asr_stats_raw_en_word/logdir/valid.1.scp
+batch_type: folded
+valid_batch_type: null
+fold_length: []
+sort_in_batch: descending
+sort_batch: descending
+multiple_iterator: false
+chunk_length: 500
+chunk_shift_ratio: 0.5
+num_cache_chunks: 1024
+train_data_path_and_name_and_type:
+-   - dump/raw/train/wav.scp
+    - speech
+    - sound
+-   - dump/raw/train/text
+    - text
+    - text
+valid_data_path_and_name_and_type:
+-   - dump/raw/valid/wav.scp
+    - speech
+    - sound
+-   - dump/raw/valid/text
+    - text
+    - text
+allow_variable_data_keys: false
+max_cache_size: 0.0
+max_cache_fd: 32
+valid_max_cache_size: null
+optim: adam
+optim_conf:
+    lr: 0.0002
+scheduler: warmuplr
+scheduler_conf:
+    warmup_steps: 25000
+token_list:
+- <blank>
+- <unk>
+- ්
+- න
+- ම
+- ක
+- ල
+- ි
+- ු
+- ග
+- ේ
+- ර
+- ත
+- ද
+- ව
+- ට
+- ඕ
+- ී
+- ප
+- ය
+- ෙ
+- ස
+- ණ
+- ා
+- ැ
+- RequestAcc.balance
+- Moneywithdraw
+- Moneydeposit
+- Moneytransfer
+- Billpayments
+- බ
+- ඉ
+- ශ
+- ෂ
+- ඩ
+- Creditcardpayments
+- එ
+- ‍
+- හ
+- <sos/eos>
+init: null
+input_size: null
+ctc_conf:
+    dropout_rate: 0.0
+    ctc_type: builtin
+    reduce: true
+    ignore_nan_grad: true
+model_conf:
+    ctc_weight: 0.5
+    ignore_id: -1
+    lsm_weight: 0.0
+    length_normalized_loss: false
+    report_cer: true
+    report_wer: true
+    sym_space: <space>
+    sym_blank: <blank>
+    extract_feats_in_collect_stats: true
+use_preprocessor: true
+token_type: word
+bpemodel: null
+non_linguistic_symbols: null
+cleaner: null
+g2p: null
+speech_volume_normalize: null
+rir_scp: null
+rir_apply_prob: 1.0
+noise_scp: null
+noise_apply_prob: 1.0
+noise_db_range: '13_15'
+frontend: default
+frontend_conf:
+    fs: 16k
+specaug: specaug
+specaug_conf:
+    apply_time_warp: true
+    time_warp_window: 5
+    time_warp_mode: bicubic
+    apply_freq_mask: true
+    freq_mask_width_range:
+    - 0
+    - 30
+    num_freq_mask: 2
+    apply_time_mask: true
+    time_mask_width_range:
+    - 0
+    - 40
+    num_time_mask: 2
+normalize: utterance_mvn
+normalize_conf: {}
+preencoder: null
+preencoder_conf: {}
+encoder: transformer
+encoder_conf:
+    output_size: 256
+    attention_heads: 4
+    linear_units: 2048
+    num_blocks: 12
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    attention_dropout_rate: 0.0
+    input_layer: conv2d
+    normalize_before: true
+postencoder: null
+postencoder_conf: {}
+decoder: transformer
+decoder_conf:
+    attention_heads: 4
+    linear_units: 2048
+    num_blocks: 6
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    self_attention_dropout_rate: 0.0
+    src_attention_dropout_rate: 0.0
+required:
+- output_dir
+- token_list
+version: 0.10.3a3
+distributed: false

exp/asr_stats_raw_en_word/logdir/stats.1/train/batch_keys

@@ -0,0 +1,2 @@
+speech
+text

exp/asr_stats_raw_en_word/logdir/stats.1/train/speech_shape

@@ -0,0 +1,191 @@
+wavs_audio1520836294077_1_1.wav 35520
+wavs_audio1520836461764_1_8.wav 48960
+wavs_audio1520837122917_1_1.wav 40320
+wavs_audio1520840663051_1_1.wav 34560
+wavs_audio1520841304669_1_1.wav 32640
+wavs_audio1520842275873_6_2.wav 44160
+wavs_audio1520843492200_2_5.wav 57600
+wavs_audio1520843508056_2_7.wav 46080
+wavs_audio1520872925842_2_2.wav 99840
+wavs_audio1521963566571_4_4.wav 53760
+wavs_audio1521963584328_4_5.wav 50880
+wavs_audio1521963604100_4_1.wav 44160
+wavs_audio1521963620973_6_3.wav 43200
+wavs_audio1521965219757_1_1.wav 111360
+wavs_audio1521967020188_1_1.wav 57344
+wavs_audio1521967072918_1_3.wav 52776
+wavs_audio1521967088867_1_4.wav 55655
+wavs_audio1521967143467_4_4.wav 37358
+wavs_audio1521967174048_5_2.wav 57344
+wavs_audio1521967182568_5_3.wav 57344
+wavs_audio1521967191647_5_4.wav 57002
+wavs_audio1521967210928_2_1.wav 50315
+wavs_audio1521967218948_2_2.wav 48086
+wavs_audio1521967234607_2_4.wav 47621
+wavs_audio1521971805599_1_2.wav 80640
+wavs_audio1521971823304_1_3.wav 48960
+wavs_audio1521971839606_1_4.wav 65280
+wavs_audio1521971851847_1_5.wav 60480
+wavs_audio1521971860971_1_6.wav 42240
+wavs_audio1521971875937_1_7.wav 69120
+wavs_audio1521971898008_1_1.wav 61440
+wavs_audio1521971912791_4_1.wav 45120
+wavs_audio1521971921691_4_2.wav 54720
+wavs_audio1521971930457_4_3.wav 45120
+wavs_audio1521972707739_4_4.wav 44160
+wavs_audio1521972722300_4_3.wav 36480
+wavs_audio1521972738905_4_5.wav 41280
+wavs_audio1521977707787_1_2.wav 53760
+wavs_audio1521977720120_1_3.wav 44160
+wavs_audio1521977752026_1_4.wav 50880
+wavs_audio1521977773902_1_5.wav 48000
+wavs_audio1521977783661_1_6.wav 48960
+wavs_audio1521977799200_1_7.wav 69120
+wavs_audio1521977813700_1_8.wav 58560
+wavs_audio1521977891866_2_5.wav 47040
+wavs_audio1521977899927_2_6.wav 50880
+wavs_audio1521977907088_2_7.wav 46080
+wavs_audio1521977947642_3_1.wav 43200
+wavs_audio1521977957590_3_2.wav 35520
+wavs_audio1521977976433_3_4.wav 36480
+wavs_audio1521977987573_3_5.wav 58560
+wavs_audio1521977996513_3_6.wav 46080
+wavs_audio1521978015573_3_8.wav 40320
+wavs_audio1521978052852_4_1.wav 45120
+wavs_audio1521978062932_4_2.wav 46080
+wavs_audio1521978075070_4_3.wav 48960
+wavs_audio1521978082126_4_4.wav 31680
+wavs_audio1521978092090_4_5.wav 46080
+wavs_audio1521978127326_5_1.wav 57600
+wavs_audio1521978179622_5_2.wav 61440
+wavs_audio1521978202372_5_3.wav 55680
+wavs_audio1521978222792_5_4.wav 61440
+wavs_audio1521978235071_5_5.wav 57600
+wavs_audio1521978250135_5_6.wav 62400
+wavs_audio1521978326183_6_4.wav 58560
+wavs_audio1521978343427_6_3.wav 54720
+wavs_audio1521978355588_6_4.wav 57600
+wavs_audio1521978677087_1_2.wav 86400
+wavs_audio1521978712263_1_1.wav 71040
+wavs_audio1521978727285_1_3.wav 48000
+wavs_audio1521978740460_1_4.wav 57600
+wavs_audio1521978758646_1_6.wav 37440
+wavs_audio1521978773477_1_7.wav 68160
+wavs_audio1521978806858_2_1.wav 48000
+wavs_audio1521978814597_2_2.wav 47040
+wavs_audio1521978821952_2_3.wav 43200
+wavs_audio1521978845653_2_6.wav 37440
+wavs_audio1521978853183_2_7.wav 45120
+wavs_audio1521978869235_1_1.wav 64320
+wavs_audio1521978872384_3_1.wav 38400
+wavs_audio1521978888019_3_3.wav 39360
+wavs_audio1521978888670_1_2.wav 50880
+wavs_audio1521978908272_3_5.wav 48000
+wavs_audio1521978914741_3_6.wav 50880
+wavs_audio1521978924267_3_7.wav 34560
+wavs_audio1521978933309_3_8.wav 35520
+wavs_audio1521978940490_1_4.wav 46080
+wavs_audio1521978950982_4_2.wav 43200
+wavs_audio1521978958260_4_3.wav 39360
+wavs_audio1521978969477_4_4.wav 39360
+wavs_audio1521978973155_1_6.wav 40320
+wavs_audio1521978993428_1_2.wav 72000
+wavs_audio1521979014107_5_1.wav 67200
+wavs_audio1521979015035_1_3.wav 58560
+wavs_audio1521979027700_5_2.wav 88320
+wavs_audio1521979039812_1_4.wav 73920
+wavs_audio1521979050276_5_4.wav 64320
+wavs_audio1521979057273_1_5.wav 69120
+wavs_audio1521979062158_5_5.wav 64320
+wavs_audio1521979078672_5_7.wav 53760
+wavs_audio1521979090759_1_3.wav 46080
+wavs_audio1521979096579_6_1.wav 51840
+wavs_audio1521979102002_1_7.wav 84480
+wavs_audio1521979106195_6_2.wav 55680
+wavs_audio1521979106979_1_4.wav 48960
+wavs_audio1521979119723_1_8.wav 95040
+wavs_audio1521979120029_1_5.wav 41280
+wavs_audio1521979123293_6_3.wav 57600
+wavs_audio1521979151139_1_7.wav 66240
+wavs_audio1521979371025_1_1.wav 64320
+wavs_audio1521979395109_1_3.wav 44160
+wavs_audio1521979407130_1_4.wav 53760
+wavs_audio1521979418969_1_5.wav 45120
+wavs_audio1521979451100_1_7.wav 77760
+wavs_audio1521979471524_1_8.wav 70080
+wavs_audio1521979769345_1_1.wav 68160
+wavs_audio1521979802399_1_2.wav 96960
+wavs_audio1521979893089_1_5.wav 83520
+wavs_audio1521979928067_1_7.wav 81600
+wavs_audio1521979952404_1_8.wav 75840
+wavs_audio1521980004898_2_1.wav 72000
+wavs_audio1521980018398_2_2.wav 51840
+wavs_audio1521980039252_2_3.wav 52800
+wavs_audio1521980051832_2_4.wav 55680
+wavs_audio1521980067912_2_5.wav 49920
+wavs_audio1521980094101_2_7.wav 53760
+wavs_audio1521980170026_3_3.wav 49920
+wavs_audio1521980191245_3_5.wav 43200
+wavs_audio1521980205044_3_6.wav 42240
+wavs_audio1521980233059_3_7.wav 89280
+wavs_audio1521980244453_3_8.wav 48960
+wavs_audio1521980278845_4_1.wav 46080
+wavs_audio1521980295672_4_2.wav 65280
+wavs_audio1521980306831_4_3.wav 48960
+wavs_audio1521980313648_4_4.wav 33600
+wavs_audio1521980348852_5_1.wav 70080
+wavs_audio1521980361431_5_2.wav 68160
+wavs_audio1521980389212_5_4.wav 69120
+wavs_audio1521980398806_5_5.wav 72960
+wavs_audio1521980413007_5_6.wav 82560
+wavs_audio1521980461538_6_1.wav 52800
+wavs_audio1521980474318_6_2.wav 72000
+wavs_audio1521985594389_3_1.wav 80640
+wavs_audio1521985613538_3_1.wav 90240
+wavs_audio1521985626118_3_1.wav 72960
+wavs_audio1521985663610_2_4.wav 82560
+wavs_audio1521985688421_2_6.wav 83520
+wavs_audio1521990339369_1_1.wav 57344
+wavs_audio1521990360269_1_2.wav 57344
+wavs_audio1521990394150_1_3.wav 57344
+wavs_audio1521990423934_1_4.wav 57344
+wavs_audio1521990453497_1_5.wav 51645
+wavs_audio1521990475391_1_6.wav 57344
+wavs_audio1521990548847_1_8.wav 57344
+wavs_audio1521990572405_2_1.wav 57344
+wavs_audio1521990606637_2_2.wav 57344
+wavs_audio1521990622647_2_3.wav 57344
+wavs_audio1521990651415_2_4.wav 57344
+wavs_audio1521990713296_2_6.wav 54973
+wavs_audio1521990733173_2_7.wav 57344
+wavs_audio1521990844646_3_3.wav 54504
+wavs_audio1521990928443_3_5.wav 57344
+wavs_audio1521990948913_3_6.wav 57344
+wavs_audio1521990965351_3_7.wav 52413
+wavs_audio1521990987567_3_8.wav 57344
+wavs_audio1521991035895_4_1.wav 45544
+wavs_audio1521991054817_4_2.wav 57344
+wavs_audio1521991153583_4_4.wav 45245
+wavs_audio1521991191047_4_5.wav 57344
+wavs_audio1521991250215_5_1.wav 57344
+wavs_audio1521991440896_5_3.wav 57344
+wavs_audio1521991611232_5_5.wav 57344
+wavs_audio1521991643172_5_6.wav 57344
+wavs_audio1521991717586_6_1.wav 57344
+wavs_audio1521991757232_6_2.wav 57344
+wavs_audio1521991785946_6_3.wav 57344
+wavs_audio1521991849774_6_4.wav 57344
+wavs_audio1521998390206_1_1.wav 48000
+wavs_audio1521998665288_2_7.wav 50880
+wavs_audio1521998752944_3_6.wav 59520
+wavs_audio1521998843131_4_4.wav 37440
+wavs_audio1521998905583_5_6.wav 51840
+wavs_audio1521998964991_6_3.wav 58560
+wavs_audio1522007605101_1_1.wav 53760
+wavs_audio1522007643961_1_3.wav 40320
+wavs_audio1522007658801_1_4.wav 40320
+wavs_audio1522007670811_1_5.wav 38400
+wavs_audio1522007681976_1_6.wav 42240
+wavs_audio1522007696866_1_7.wav 50880
+wavs_audio1522007710971_1_8.wav 49920
+wavs_audio1522007735606_1_7.wav 87360

exp/asr_stats_raw_en_word/logdir/stats.1/train/stats_keys

@@ -0,0 +1,2 @@
+feats
+feats_lengths

exp/asr_stats_raw_en_word/logdir/stats.1/train/text_shape

@@ -0,0 +1,191 @@
+wavs_audio1520836294077_1_1.wav 18
+wavs_audio1520836461764_1_8.wav 33
+wavs_audio1520837122917_1_1.wav 18
+wavs_audio1520840663051_1_1.wav 18
+wavs_audio1520841304669_1_1.wav 18
+wavs_audio1520842275873_6_2.wav 23
+wavs_audio1520843492200_2_5.wav 14
+wavs_audio1520843508056_2_7.wav 20
+wavs_audio1520872925842_2_2.wav 23
+wavs_audio1521963566571_4_4.wav 12
+wavs_audio1521963584328_4_5.wav 20
+wavs_audio1521963604100_4_1.wav 18
+wavs_audio1521963620973_6_3.wav 33
+wavs_audio1521965219757_1_1.wav 18
+wavs_audio1521967020188_1_1.wav 18
+wavs_audio1521967072918_1_3.wav 9
+wavs_audio1521967088867_1_4.wav 20
+wavs_audio1521967143467_4_4.wav 12
+wavs_audio1521967174048_5_2.wav 32
+wavs_audio1521967182568_5_3.wav 29
+wavs_audio1521967191647_5_4.wav 29
+wavs_audio1521967210928_2_1.wav 23
+wavs_audio1521967218948_2_2.wav 23
+wavs_audio1521967234607_2_4.wav 21
+wavs_audio1521971805599_1_2.wav 15
+wavs_audio1521971823304_1_3.wav 9
+wavs_audio1521971839606_1_4.wav 20
+wavs_audio1521971851847_1_5.wav 17
+wavs_audio1521971860971_1_6.wav 11
+wavs_audio1521971875937_1_7.wav 31
+wavs_audio1521971898008_1_1.wav 18
+wavs_audio1521971912791_4_1.wav 18
+wavs_audio1521971921691_4_2.wav 20
+wavs_audio1521971930457_4_3.wav 18
+wavs_audio1521972707739_4_4.wav 12
+wavs_audio1521972722300_4_3.wav 18
+wavs_audio1521972738905_4_5.wav 20
+wavs_audio1521977707787_1_2.wav 15
+wavs_audio1521977720120_1_3.wav 9
+wavs_audio1521977752026_1_4.wav 20
+wavs_audio1521977773902_1_5.wav 17
+wavs_audio1521977783661_1_6.wav 11
+wavs_audio1521977799200_1_7.wav 31
+wavs_audio1521977813700_1_8.wav 33
+wavs_audio1521977891866_2_5.wav 14
+wavs_audio1521977899927_2_6.wav 20
+wavs_audio1521977907088_2_7.wav 20
+wavs_audio1521977947642_3_1.wav 15
+wavs_audio1521977957590_3_2.wav 13
+wavs_audio1521977976433_3_4.wav 13
+wavs_audio1521977987573_3_5.wav 20
+wavs_audio1521977996513_3_6.wav 20
+wavs_audio1521978015573_3_8.wav 13
+wavs_audio1521978052852_4_1.wav 18
+wavs_audio1521978062932_4_2.wav 20
+wavs_audio1521978075070_4_3.wav 18
+wavs_audio1521978082126_4_4.wav 12
+wavs_audio1521978092090_4_5.wav 20
+wavs_audio1521978127326_5_1.wav 27
+wavs_audio1521978179622_5_2.wav 32
+wavs_audio1521978202372_5_3.wav 29
+wavs_audio1521978222792_5_4.wav 29
+wavs_audio1521978235071_5_5.wav 27
+wavs_audio1521978250135_5_6.wav 27
+wavs_audio1521978326183_6_4.wav 35
+wavs_audio1521978343427_6_3.wav 33
+wavs_audio1521978355588_6_4.wav 35
+wavs_audio1521978677087_1_2.wav 15
+wavs_audio1521978712263_1_1.wav 18
+wavs_audio1521978727285_1_3.wav 9
+wavs_audio1521978740460_1_4.wav 20
+wavs_audio1521978758646_1_6.wav 11
+wavs_audio1521978773477_1_7.wav 31
+wavs_audio1521978806858_2_1.wav 23
+wavs_audio1521978814597_2_2.wav 23
+wavs_audio1521978821952_2_3.wav 21
+wavs_audio1521978845653_2_6.wav 20
+wavs_audio1521978853183_2_7.wav 20
+wavs_audio1521978869235_1_1.wav 18
+wavs_audio1521978872384_3_1.wav 15
+wavs_audio1521978888019_3_3.wav 15
+wavs_audio1521978888670_1_2.wav 15
+wavs_audio1521978908272_3_5.wav 20
+wavs_audio1521978914741_3_6.wav 20
+wavs_audio1521978924267_3_7.wav 13
+wavs_audio1521978933309_3_8.wav 13
+wavs_audio1521978940490_1_4.wav 20
+wavs_audio1521978950982_4_2.wav 20
+wavs_audio1521978958260_4_3.wav 18
+wavs_audio1521978969477_4_4.wav 12
+wavs_audio1521978973155_1_6.wav 11
+wavs_audio1521978993428_1_2.wav 15
+wavs_audio1521979014107_5_1.wav 27
+wavs_audio1521979015035_1_3.wav 9
+wavs_audio1521979027700_5_2.wav 32
+wavs_audio1521979039812_1_4.wav 20
+wavs_audio1521979050276_5_4.wav 29
+wavs_audio1521979057273_1_5.wav 17
+wavs_audio1521979062158_5_5.wav 27
+wavs_audio1521979078672_5_7.wav 29
+wavs_audio1521979090759_1_3.wav 9
+wavs_audio1521979096579_6_1.wav 21
+wavs_audio1521979102002_1_7.wav 31
+wavs_audio1521979106195_6_2.wav 23
+wavs_audio1521979106979_1_4.wav 20
+wavs_audio1521979119723_1_8.wav 33
+wavs_audio1521979120029_1_5.wav 17
+wavs_audio1521979123293_6_3.wav 33
+wavs_audio1521979151139_1_7.wav 31
+wavs_audio1521979371025_1_1.wav 18
+wavs_audio1521979395109_1_3.wav 9
+wavs_audio1521979407130_1_4.wav 20
+wavs_audio1521979418969_1_5.wav 17
+wavs_audio1521979451100_1_7.wav 31
+wavs_audio1521979471524_1_8.wav 33
+wavs_audio1521979769345_1_1.wav 18
+wavs_audio1521979802399_1_2.wav 15
+wavs_audio1521979893089_1_5.wav 17
+wavs_audio1521979928067_1_7.wav 31
+wavs_audio1521979952404_1_8.wav 33
+wavs_audio1521980004898_2_1.wav 23
+wavs_audio1521980018398_2_2.wav 23
+wavs_audio1521980039252_2_3.wav 21
+wavs_audio1521980051832_2_4.wav 21
+wavs_audio1521980067912_2_5.wav 14
+wavs_audio1521980094101_2_7.wav 20
+wavs_audio1521980170026_3_3.wav 15
+wavs_audio1521980191245_3_5.wav 20
+wavs_audio1521980205044_3_6.wav 20
+wavs_audio1521980233059_3_7.wav 13
+wavs_audio1521980244453_3_8.wav 13
+wavs_audio1521980278845_4_1.wav 18
+wavs_audio1521980295672_4_2.wav 20
+wavs_audio1521980306831_4_3.wav 18
+wavs_audio1521980313648_4_4.wav 12
+wavs_audio1521980348852_5_1.wav 27
+wavs_audio1521980361431_5_2.wav 32
+wavs_audio1521980389212_5_4.wav 29
+wavs_audio1521980398806_5_5.wav 27
+wavs_audio1521980413007_5_6.wav 27
+wavs_audio1521980461538_6_1.wav 21
+wavs_audio1521980474318_6_2.wav 23
+wavs_audio1521985594389_3_1.wav 15
+wavs_audio1521985613538_3_1.wav 15
+wavs_audio1521985626118_3_1.wav 15
+wavs_audio1521985663610_2_4.wav 21
+wavs_audio1521985688421_2_6.wav 20
+wavs_audio1521990339369_1_1.wav 18
+wavs_audio1521990360269_1_2.wav 15
+wavs_audio1521990394150_1_3.wav 9
+wavs_audio1521990423934_1_4.wav 20
+wavs_audio1521990453497_1_5.wav 17
+wavs_audio1521990475391_1_6.wav 11
+wavs_audio1521990548847_1_8.wav 33
+wavs_audio1521990572405_2_1.wav 23
+wavs_audio1521990606637_2_2.wav 23
+wavs_audio1521990622647_2_3.wav 21
+wavs_audio1521990651415_2_4.wav 21
+wavs_audio1521990713296_2_6.wav 20
+wavs_audio1521990733173_2_7.wav 20
+wavs_audio1521990844646_3_3.wav 15
+wavs_audio1521990928443_3_5.wav 20
+wavs_audio1521990948913_3_6.wav 20
+wavs_audio1521990965351_3_7.wav 13
+wavs_audio1521990987567_3_8.wav 13
+wavs_audio1521991035895_4_1.wav 18
+wavs_audio1521991054817_4_2.wav 20
+wavs_audio1521991153583_4_4.wav 12
+wavs_audio1521991191047_4_5.wav 20
+wavs_audio1521991250215_5_1.wav 27
+wavs_audio1521991440896_5_3.wav 29
+wavs_audio1521991611232_5_5.wav 27
+wavs_audio1521991643172_5_6.wav 27
+wavs_audio1521991717586_6_1.wav 21
+wavs_audio1521991757232_6_2.wav 23
+wavs_audio1521991785946_6_3.wav 33
+wavs_audio1521991849774_6_4.wav 35
+wavs_audio1521998390206_1_1.wav 18
+wavs_audio1521998665288_2_7.wav 20
+wavs_audio1521998752944_3_6.wav 20
+wavs_audio1521998843131_4_4.wav 12
+wavs_audio1521998905583_5_6.wav 27
+wavs_audio1521998964991_6_3.wav 33
+wavs_audio1522007605101_1_1.wav 18
+wavs_audio1522007643961_1_3.wav 9
+wavs_audio1522007658801_1_4.wav 20
+wavs_audio1522007670811_1_5.wav 17
+wavs_audio1522007681976_1_6.wav 11
+wavs_audio1522007696866_1_7.wav 31
+wavs_audio1522007710971_1_8.wav 33
+wavs_audio1522007735606_1_7.wav 31

exp/asr_stats_raw_en_word/logdir/stats.1/valid/batch_keys

@@ -0,0 +1,2 @@
+speech
+text

exp/asr_stats_raw_en_word/logdir/stats.1/valid/speech_shape

@@ -0,0 +1,24 @@
+wavs_audio1521967108757_4_1.wav 48922
+wavs_audio1521967226793_2_3.wav 49340
+wavs_audio1521971937930_4_4.wav 38400
+wavs_audio1521971946892_4_5.wav 44160
+wavs_audio1521977862147_2_2.wav 53760
+wavs_audio1521978750025_1_5.wav 46080
+wavs_audio1521978830797_2_4.wav 44160
+wavs_audio1521978837483_2_5.wav 42240
+wavs_audio1521978880668_3_2.wav 37440
+wavs_audio1521978977408_4_5.wav 48000
+wavs_audio1521978992321_4_1.wav 47040
+wavs_audio1521979002585_1_8.wav 52800
+wavs_audio1521979070192_5_6.wav 57600
+wavs_audio1521979075133_1_6.wav 59520
+wavs_audio1521979077899_1_2.wav 49920
+wavs_audio1521979132984_1_6.wav 38400
+wavs_audio1521979147057_6_4.wav 65280
+wavs_audio1521979384490_1_2.wav 52800
+wavs_audio1521979852020_1_4.wav 101760
+wavs_audio1521980432472_5_7.wav 71040
+wavs_audio1521980519079_6_4.wav 70080
+wavs_audio1521986099502_1_6.wav 94080
+wavs_audio1521990528827_1_7.wav 57344
+wavs_audio1521990793862_3_2.wav 38973

exp/asr_stats_raw_en_word/logdir/stats.1/valid/stats_keys

@@ -0,0 +1,2 @@
+feats
+feats_lengths

exp/asr_stats_raw_en_word/logdir/stats.1/valid/text_shape

@@ -0,0 +1,24 @@
+wavs_audio1521967108757_4_1.wav 18
+wavs_audio1521967226793_2_3.wav 21
+wavs_audio1521971937930_4_4.wav 12
+wavs_audio1521971946892_4_5.wav 20
+wavs_audio1521977862147_2_2.wav 23
+wavs_audio1521978750025_1_5.wav 17
+wavs_audio1521978830797_2_4.wav 21
+wavs_audio1521978837483_2_5.wav 14
+wavs_audio1521978880668_3_2.wav 13
+wavs_audio1521978977408_4_5.wav 20
+wavs_audio1521978992321_4_1.wav 18
+wavs_audio1521979002585_1_8.wav 33
+wavs_audio1521979070192_5_6.wav 27
+wavs_audio1521979075133_1_6.wav 11
+wavs_audio1521979077899_1_2.wav 15
+wavs_audio1521979132984_1_6.wav 11
+wavs_audio1521979147057_6_4.wav 35
+wavs_audio1521979384490_1_2.wav 15
+wavs_audio1521979852020_1_4.wav 20
+wavs_audio1521980432472_5_7.wav 29
+wavs_audio1521980519079_6_4.wav 35
+wavs_audio1521986099502_1_6.wav 11
+wavs_audio1521990528827_1_7.wav 31
+wavs_audio1521990793862_3_2.wav 13

exp/asr_stats_raw_en_word/logdir/stats.10.log

@@ -0,0 +1,495 @@
+# Running on r099.ib.bridges2.psc.edu
+# Started at Tue Dec 21 22:24:49 EST 2021
+# SLURMD_NODENAME=r099
+# SLURM_ARRAY_JOB_ID=5730432
+# SLURM_ARRAY_TASK_COUNT=32
+# SLURM_ARRAY_TASK_ID=10
+# SLURM_ARRAY_TASK_MAX=32
+# SLURM_ARRAY_TASK_MIN=1
+# SLURM_ARRAY_TASK_STEP=1
+# SLURM_CLUSTER_NAME=bridges2
+# SLURM_CONF=/var/spool/slurm/d/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=1
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GTIDS=0
+# SLURM_JOBID=5730463
+# SLURM_JOB_ACCOUNT=cis210027p
+# SLURM_JOB_CPUS_PER_NODE=1
+# SLURM_JOB_GID=24886
+# SLURM_JOB_ID=5730463
+# SLURM_JOB_NAME=stats.sh
+# SLURM_JOB_NODELIST=r099
+# SLURM_JOB_NUM_NODES=1
+# SLURM_JOB_PARTITION=RM-shared
+# SLURM_JOB_QOS=rm
+# SLURM_JOB_UID=82326
+# SLURM_JOB_USER=ganesank
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_CPU=2000
+# SLURM_NNODES=1
+# SLURM_NODEID=0
+# SLURM_NODELIST=r099
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/ocean/projects/cis210027p/ganesank/karthik_new/espnet/egs2/sinhala/asr1
+# SLURM_SUBMIT_HOST=br012.ib.bridges2.psc.edu
+# SLURM_TASKS_PER_NODE=1
+# SLURM_TASK_PID=6007
+# SLURM_TOPOLOGY_ADDR=r099
+# SLURM_TOPOLOGY_ADDR_PATTERN=node
+# SLURM_WORKING_CLUSTER=bridges2:br003:6814:9216:109
+# python3 -m espnet2.bin.asr_train --collect_stats true --use_preprocessor true --bpemodel none --token_type word --token_list data/en_token_list/word/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --train_data_path_and_name_and_type dump/raw/train/wav.scp,speech,sound --train_data_path_and_name_and_type dump/raw/train/text,text,text --valid_data_path_and_name_and_type dump/raw/valid/wav.scp,speech,sound --valid_data_path_and_name_and_type dump/raw/valid/text,text,text --train_shape_file exp/asr_stats_raw_en_word/logdir/train.10.scp --valid_shape_file exp/asr_stats_raw_en_word/logdir/valid.10.scp --output_dir exp/asr_stats_raw_en_word/logdir/stats.10 --config conf/train_asr.yaml --frontend_conf fs=16k 
+/ocean/projects/cis210027p/ganesank/karthik_new/espnet/tools/venv/bin/python3 /ocean/projects/cis210027p/ganesank/karthik_new/espnet/espnet2/bin/asr_train.py --collect_stats true --use_preprocessor true --bpemodel none --token_type word --token_list data/en_token_list/word/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --train_data_path_and_name_and_type dump/raw/train/wav.scp,speech,sound --train_data_path_and_name_and_type dump/raw/train/text,text,text --valid_data_path_and_name_and_type dump/raw/valid/wav.scp,speech,sound --valid_data_path_and_name_and_type dump/raw/valid/text,text,text --train_shape_file exp/asr_stats_raw_en_word/logdir/train.10.scp --valid_shape_file exp/asr_stats_raw_en_word/logdir/valid.10.scp --output_dir exp/asr_stats_raw_en_word/logdir/stats.10 --config conf/train_asr.yaml --frontend_conf fs=16k
+[r099] 2021-12-21 22:25:08,042 (asr:382) INFO: Vocabulary size: 40
+[r099] 2021-12-21 22:25:09,019 (abs_task:1132) INFO: pytorch.version=1.8.1+cu102, cuda.available=False, cudnn.version=7605, cudnn.benchmark=False, cudnn.deterministic=True
+[r099] 2021-12-21 22:25:09,024 (abs_task:1133) INFO: Model structure:
+ESPnetASRModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=512, hop_length=128, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (time_warp): TimeWarp(window=5, mode=bicubic)
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 30], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxis(mask_width_range=[0, 40], num_mask=2, axis=time)
+  )
+  (normalize): UtteranceMVN(norm_means=True, norm_vars=False)
+  (encoder): TransformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 256, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=4864, out_features=256, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (9): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (10): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (11): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(40, 256)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=256, out_features=40, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=256, out_features=40, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetASRModel
+    Total Number of model parameters: 27.12 M
+    Number of trainable parameters: 27.12 M (100.0%)
+    Size: 108.49 MB
+    Type: torch.float32
+[r099] 2021-12-21 22:25:09,024 (abs_task:1136) INFO: Optimizer:
+Adam (
+Parameter Group 0
+    amsgrad: False
+    betas: (0.9, 0.999)
+    eps: 1e-08
+    initial_lr: 0.0002
+    lr: 8e-09
+    weight_decay: 0
+)
+[r099] 2021-12-21 22:25:09,024 (abs_task:1137) INFO: Scheduler: WarmupLR(warmup_steps=25000)
+[r099] 2021-12-21 22:25:09,027 (abs_task:1146) INFO: Saving the configuration in exp/asr_stats_raw_en_word/logdir/stats.10/config.yaml
+[r099] 2021-12-21 22:25:09,037 (abs_task:1157) INFO: Namespace(config='conf/train_asr.yaml', print_config=False, log_level='INFO', dry_run=False, iterator_type='sequence', output_dir='exp/asr_stats_raw_en_word/logdir/stats.10', ngpu=0, seed=0, num_workers=1, num_att_plot=3, dist_backend='nccl', dist_init_method='env://', dist_world_size=None, dist_rank=None, local_rank=None, dist_master_addr=None, dist_master_port=None, dist_launcher=None, multiprocessing_distributed=False, unused_parameters=False, sharded_ddp=False, cudnn_enabled=True, cudnn_benchmark=False, cudnn_deterministic=True, collect_stats=True, write_collected_feats=False, max_epoch=50, patience=None, val_scheduler_criterion=('valid', 'loss'), early_stopping_criterion=('valid', 'loss', 'min'), best_model_criterion=[('train', 'loss', 'min'), ('valid', 'loss', 'min'), ('train', 'acc', 'max'), ('valid', 'acc', 'max')], keep_nbest_models=5, grad_clip=5.0, grad_clip_type=2.0, grad_noise=False, accum_grad=1, no_forward_run=False, resume=False, train_dtype='float32', use_amp=False, log_interval=None, use_tensorboard=True, use_wandb=False, wandb_project=None, wandb_id=None, wandb_entity=None, wandb_name=None, wandb_model_log_interval=-1, detect_anomaly=False, pretrain_path=None, init_param=[], ignore_init_mismatch=False, freeze_param=[], num_iters_per_epoch=None, batch_size=20, valid_batch_size=None, batch_bins=1000000, valid_batch_bins=None, train_shape_file=['exp/asr_stats_raw_en_word/logdir/train.10.scp'], valid_shape_file=['exp/asr_stats_raw_en_word/logdir/valid.10.scp'], batch_type='folded', valid_batch_type=None, fold_length=[], sort_in_batch='descending', sort_batch='descending', multiple_iterator=False, chunk_length=500, chunk_shift_ratio=0.5, num_cache_chunks=1024, train_data_path_and_name_and_type=[('dump/raw/train/wav.scp', 'speech', 'sound'), ('dump/raw/train/text', 'text', 'text')], valid_data_path_and_name_and_type=[('dump/raw/valid/wav.scp', 'speech', 'sound'), ('dump/raw/valid/text', 'text', 'text')], allow_variable_data_keys=False, max_cache_size=0.0, max_cache_fd=32, valid_max_cache_size=None, optim='adam', optim_conf={'lr': 0.0002}, scheduler='warmuplr', scheduler_conf={'warmup_steps': 25000}, token_list=['<blank>', '<unk>', '්', 'න', 'ම', 'ක', 'ල', 'ි', 'ු', 'ග', 'ේ', 'ර', 'ත', 'ද', 'ව', 'ට', 'ඕ', 'ී', 'ප', 'ය', 'ෙ', 'ස', 'ණ', 'ා', 'ැ', 'RequestAcc.balance', 'Moneywithdraw', 'Moneydeposit', 'Moneytransfer', 'Billpayments', 'බ', 'ඉ', 'ශ', 'ෂ', 'ඩ', 'Creditcardpayments', 'එ', '\u200d', 'හ', '<sos/eos>'], init=None, input_size=None, ctc_conf={'dropout_rate': 0.0, 'ctc_type': 'builtin', 'reduce': True, 'ignore_nan_grad': True}, model_conf={'ctc_weight': 0.5, 'ignore_id': -1, 'lsm_weight': 0.0, 'length_normalized_loss': False, 'report_cer': True, 'report_wer': True, 'sym_space': '<space>', 'sym_blank': '<blank>', 'extract_feats_in_collect_stats': True}, use_preprocessor=True, token_type='word', bpemodel=None, non_linguistic_symbols=None, cleaner=None, g2p=None, speech_volume_normalize=None, rir_scp=None, rir_apply_prob=1.0, noise_scp=None, noise_apply_prob=1.0, noise_db_range='13_15', frontend='default', frontend_conf={'fs': '16k'}, specaug='specaug', specaug_conf={'apply_time_warp': True, 'time_warp_window': 5, 'time_warp_mode': 'bicubic', 'apply_freq_mask': True, 'freq_mask_width_range': [0, 30], 'num_freq_mask': 2, 'apply_time_mask': True, 'time_mask_width_range': [0, 40], 'num_time_mask': 2}, normalize='utterance_mvn', normalize_conf={}, preencoder=None, preencoder_conf={}, encoder='transformer', encoder_conf={'output_size': 256, 'attention_heads': 4, 'linear_units': 2048, 'num_blocks': 12, 'dropout_rate': 0.1, 'positional_dropout_rate': 0.1, 'attention_dropout_rate': 0.0, 'input_layer': 'conv2d', 'normalize_before': True}, postencoder=None, postencoder_conf={}, decoder='transformer', decoder_conf={'attention_heads': 4, 'linear_units': 2048, 'num_blocks': 6, 'dropout_rate': 0.1, 'positional_dropout_rate': 0.1, 'self_attention_dropout_rate': 0.0, 'src_attention_dropout_rate': 0.0}, required=['output_dir', 'token_list'], version='0.10.3a3', distributed=False)
+# Accounting: begin_time=1640143489
+# Accounting: end_time=1640143519
+# Accounting: time=30 threads=1
+# Finished at Tue Dec 21 22:25:19 EST 2021 with status 0

exp/asr_stats_raw_en_word/logdir/stats.10/config.yaml

@@ -0,0 +1,231 @@
+config: conf/train_asr.yaml
+print_config: false
+log_level: INFO
+dry_run: false
+iterator_type: sequence
+output_dir: exp/asr_stats_raw_en_word/logdir/stats.10
+ngpu: 0
+seed: 0
+num_workers: 1
+num_att_plot: 3
+dist_backend: nccl
+dist_init_method: env://
+dist_world_size: null
+dist_rank: null
+local_rank: null
+dist_master_addr: null
+dist_master_port: null
+dist_launcher: null
+multiprocessing_distributed: false
+unused_parameters: false
+sharded_ddp: false
+cudnn_enabled: true
+cudnn_benchmark: false
+cudnn_deterministic: true
+collect_stats: true
+write_collected_feats: false
+max_epoch: 50
+patience: null
+val_scheduler_criterion:
+- valid
+- loss
+early_stopping_criterion:
+- valid
+- loss
+- min
+best_model_criterion:
+-   - train
+    - loss
+    - min
+-   - valid
+    - loss
+    - min
+-   - train
+    - acc
+    - max
+-   - valid
+    - acc
+    - max
+keep_nbest_models: 5
+grad_clip: 5.0
+grad_clip_type: 2.0
+grad_noise: false
+accum_grad: 1
+no_forward_run: false
+resume: false
+train_dtype: float32
+use_amp: false
+log_interval: null
+use_tensorboard: true
+use_wandb: false
+wandb_project: null
+wandb_id: null
+wandb_entity: null
+wandb_name: null
+wandb_model_log_interval: -1
+detect_anomaly: false
+pretrain_path: null
+init_param: []
+ignore_init_mismatch: false
+freeze_param: []
+num_iters_per_epoch: null
+batch_size: 20
+valid_batch_size: null
+batch_bins: 1000000
+valid_batch_bins: null
+train_shape_file:
+- exp/asr_stats_raw_en_word/logdir/train.10.scp
+valid_shape_file:
+- exp/asr_stats_raw_en_word/logdir/valid.10.scp
+batch_type: folded
+valid_batch_type: null
+fold_length: []
+sort_in_batch: descending
+sort_batch: descending
+multiple_iterator: false
+chunk_length: 500
+chunk_shift_ratio: 0.5
+num_cache_chunks: 1024
+train_data_path_and_name_and_type:
+-   - dump/raw/train/wav.scp
+    - speech
+    - sound
+-   - dump/raw/train/text
+    - text
+    - text
+valid_data_path_and_name_and_type:
+-   - dump/raw/valid/wav.scp
+    - speech
+    - sound
+-   - dump/raw/valid/text
+    - text
+    - text
+allow_variable_data_keys: false
+max_cache_size: 0.0
+max_cache_fd: 32
+valid_max_cache_size: null
+optim: adam
+optim_conf:
+    lr: 0.0002
+scheduler: warmuplr
+scheduler_conf:
+    warmup_steps: 25000
+token_list:
+- <blank>
+- <unk>
+- ්
+- න
+- ම
+- ක
+- ල
+- ි
+- ු
+- ග
+- ේ
+- ර
+- ත
+- ද
+- ව
+- ට
+- ඕ
+- ී
+- ප
+- ය
+- ෙ
+- ස
+- ණ
+- ා
+- ැ
+- RequestAcc.balance
+- Moneywithdraw
+- Moneydeposit
+- Moneytransfer
+- Billpayments
+- බ
+- ඉ
+- ශ
+- ෂ
+- ඩ
+- Creditcardpayments
+- එ
+- ‍
+- හ
+- <sos/eos>
+init: null
+input_size: null
+ctc_conf:
+    dropout_rate: 0.0
+    ctc_type: builtin
+    reduce: true
+    ignore_nan_grad: true
+model_conf:
+    ctc_weight: 0.5
+    ignore_id: -1
+    lsm_weight: 0.0
+    length_normalized_loss: false
+    report_cer: true
+    report_wer: true
+    sym_space: <space>
+    sym_blank: <blank>
+    extract_feats_in_collect_stats: true
+use_preprocessor: true
+token_type: word
+bpemodel: null
+non_linguistic_symbols: null
+cleaner: null
+g2p: null
+speech_volume_normalize: null
+rir_scp: null
+rir_apply_prob: 1.0
+noise_scp: null
+noise_apply_prob: 1.0
+noise_db_range: '13_15'
+frontend: default
+frontend_conf:
+    fs: 16k
+specaug: specaug
+specaug_conf:
+    apply_time_warp: true
+    time_warp_window: 5
+    time_warp_mode: bicubic
+    apply_freq_mask: true
+    freq_mask_width_range:
+    - 0
+    - 30
+    num_freq_mask: 2
+    apply_time_mask: true
+    time_mask_width_range:
+    - 0
+    - 40
+    num_time_mask: 2
+normalize: utterance_mvn
+normalize_conf: {}
+preencoder: null
+preencoder_conf: {}
+encoder: transformer
+encoder_conf:
+    output_size: 256
+    attention_heads: 4
+    linear_units: 2048
+    num_blocks: 12
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    attention_dropout_rate: 0.0
+    input_layer: conv2d
+    normalize_before: true
+postencoder: null
+postencoder_conf: {}
+decoder: transformer
+decoder_conf:
+    attention_heads: 4
+    linear_units: 2048
+    num_blocks: 6
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    self_attention_dropout_rate: 0.0
+    src_attention_dropout_rate: 0.0
+required:
+- output_dir
+- token_list
+version: 0.10.3a3
+distributed: false

exp/asr_stats_raw_en_word/logdir/stats.10/train/batch_keys

@@ -0,0 +1,2 @@
+speech
+text

exp/asr_stats_raw_en_word/logdir/stats.10/train/speech_shape

@@ -0,0 +1,191 @@
+wavs_audio1526983626238_6_1.wav 66240
+wavs_audio1526983636871_5_2.wav 51840
+wavs_audio1526983641501_5_3.wav 51840
+wavs_audio1526983643559_6_2.wav 60480
+wavs_audio1526983652301_1_1.wav 56640
+wavs_audio1526983653669_5_1.wav 56640
+wavs_audio1526983663015_5_3.wav 53760
+wavs_audio1526983668420_1_2.wav 42240
+wavs_audio1526983671576_5_3.wav 51840
+wavs_audio1526983674709_5_5.wav 47040
+wavs_audio1526983682930_1_3.wav 39360
+wavs_audio1526983683850_5_6.wav 53760
+wavs_audio1526983684511_5_4.wav 51840
+wavs_audio1526983687446_6_4.wav 61440
+wavs_audio1526983694130_5_7.wav 59520
+wavs_audio1526983695453_1_4.wav 42240
+wavs_audio1526983711192_1_5.wav 44160
+wavs_audio1526983718405_5_5.wav 57600
+wavs_audio1526983722788_6_2.wav 50880
+wavs_audio1526983726631_5_6.wav 49920
+wavs_audio1526983730520_6_3.wav 51840
+wavs_audio1526983732202_1_6.wav 41280
+wavs_audio1526983741838_6_4.wav 57600
+wavs_audio1526983744462_5_7.wav 51840
+wavs_audio1526983759496_6_1.wav 48000
+wavs_audio1526983771215_6_2.wav 45120
+wavs_audio1526983778361_1_1.wav 72000
+wavs_audio1526983780570_6_3.wav 44160
+wavs_audio1526983788348_6_4.wav 52800
+wavs_audio1526983790175_2_4.wav 54720
+wavs_audio1526983800150_1_2.wav 55680
+wavs_audio1526983802451_2_5.wav 48000
+wavs_audio1526983812921_1_3.wav 44160
+wavs_audio1526983815268_2_6.wav 52800
+wavs_audio1526983824560_2_7.wav 53760
+wavs_audio1526983825156_1_4.wav 45120
+wavs_audio1526983850722_1_5.wav 48960
+wavs_audio1526983866391_1_6.wav 44160
+wavs_audio1526983882309_1_7.wav 60480
+wavs_audio1526983886714_3_1.wav 48000
+wavs_audio1526983895919_1_8.wav 54720
+wavs_audio1526983898472_3_2.wav 30720
+wavs_audio1526983911045_3_3.wav 41280
+wavs_audio1526983930113_3_4.wav 44160
+wavs_audio1526983931076_2_1.wav 54720
+wavs_audio1526983945675_2_2.wav 52800
+wavs_audio1526983972052_3_6.wav 59520
+wavs_audio1526983981691_1_3.wav 52800
+wavs_audio1526983990965_1_2.wav 57600
+wavs_audio1526983992330_3_7.wav 46080
+wavs_audio1526983996751_2_4.wav 52800
+wavs_audio1526983997319_1_2.wav 66240
+wavs_audio1526984003559_3_8.wav 39360
+wavs_audio1526984007551_2_5.wav 35520
+wavs_audio1526984011260_1_3.wav 58560
+wavs_audio1526984017034_2_6.wav 34560
+wavs_audio1526984027086_1_4.wav 59520
+wavs_audio1526984029175_4_1.wav 45120
+wavs_audio1526984031089_1_4.wav 68160
+wavs_audio1526984031785_1_1.wav 70080
+wavs_audio1526984036073_2_7.wav 45120
+wavs_audio1526984040829_1_5.wav 49920
+wavs_audio1526984044743_4_2.wav 48000
+wavs_audio1526984055641_4_3.wav 40320
+wavs_audio1526984071399_4_4.wav 44160
+wavs_audio1526984082880_4_5.wav 57600
+wavs_audio1526984083121_1_8.wav 72000
+wavs_audio1526984083509_1_7.wav 72960
+wavs_audio1526984106151_3_2.wav 37440
+wavs_audio1526984114132_5_1.wav 69120
+wavs_audio1526984121674_3_3.wav 38400
+wavs_audio1526984123594_2_4.wav 50880
+wavs_audio1526984135088_2_5.wav 29760
+wavs_audio1526984139552_3_4.wav 31680
+wavs_audio1526984144689_2_6.wav 50880
+wavs_audio1526984147192_1_1.wav 41280
+wavs_audio1526984148728_2_4.wav 72960
+wavs_audio1526984153209_2_7.wav 49920
+wavs_audio1526984156770_3_5.wav 42240
+wavs_audio1526984169153_3_6.wav 52800
+wavs_audio1526984171836_2_1.wav 55680
+wavs_audio1526984173406_2_5.wav 46080
+wavs_audio1526984173931_5_3.wav 70080
+wavs_audio1526984179972_3_1.wav 37440
+wavs_audio1526984186071_3_7.wav 39360
+wavs_audio1526984186573_2_2.wav 59520
+wavs_audio1526984197610_2_3.wav 57600
+wavs_audio1526984200561_3_2.wav 41280
+wavs_audio1526984206448_5_5.wav 65280
+wavs_audio1526984214431_3_3.wav 45120
+wavs_audio1526984217633_3_8.wav 37440
+wavs_audio1526984223567_1_1.wav 48960
+wavs_audio1526984236297_1_2.wav 43200
+wavs_audio1526984237031_3_4.wav 56640
+wavs_audio1526984241114_5_7.wav 68160
+wavs_audio1526984243264_1_4.wav 63360
+wavs_audio1526984245216_2_2.wav 72960
+wavs_audio1526984248224_1_3.wav 45120
+wavs_audio1526984249624_3_5.wav 47040
+wavs_audio1526984255903_2_5.wav 48960
+wavs_audio1526984256425_3_6.wav 42240
+wavs_audio1526984262616_2_2.wav 72000
+wavs_audio1526984265407_1_4.wav 55680
+wavs_audio1526984281593_6_1.wav 66240
+wavs_audio1526984282940_1_6.wav 56640
+wavs_audio1526984295750_2_7.wav 57600
+wavs_audio1526984296338_4_1.wav 83520
+wavs_audio1526984301655_1_6.wav 37440
+wavs_audio1526984307908_1_7.wav 83520
+wavs_audio1526984311940_3_2.wav 51840
+wavs_audio1526984318859_4_2.wav 43200
+wavs_audio1526984321249_3_1.wav 100800
+wavs_audio1526984323422_5_1.wav 86400
+wavs_audio1526984323982_4_1.wav 36480
+wavs_audio1526984326235_1_7.wav 66240
+wavs_audio1526984327222_4_3.wav 60480
+wavs_audio1526984333916_4_4.wav 32640
+wavs_audio1526984334159_4_2.wav 44160
+wavs_audio1526984337198_1_8.wav 65280
+wavs_audio1526984337797_6_3.wav 65280
+wavs_audio1526984345066_4_5.wav 66240
+wavs_audio1526984345363_4_3.wav 38400
+wavs_audio1526984349176_6_3.wav 73920
+wavs_audio1526984351152_3_3.wav 44160
+wavs_audio1526984354681_4_4.wav 29760
+wavs_audio1526984354872_2_4.wav 43200
+wavs_audio1526984357564_6_4.wav 77760
+wavs_audio1526984363716_4_5.wav 35520
+wavs_audio1526984365423_2_1.wav 47040
+wavs_audio1526984372167_3_4.wav 43200
+wavs_audio1526984375689_2_2.wav 50880
+wavs_audio1526984377956_2_1.wav 84480
+wavs_audio1526984381963_5_1.wav 49920
+wavs_audio1526984384912_3_5.wav 53760
+wavs_audio1526984392258_2_3.wav 47040
+wavs_audio1526984400068_6_4.wav 76800
+wavs_audio1526984402818_5_2.wav 51840
+wavs_audio1526984404232_5_2.wav 67200
+wavs_audio1526984407407_3_6.wav 83520
+wavs_audio1526984414330_2_6.wav 53760
+wavs_audio1526984416571_5_3.wav 51840
+wavs_audio1526984417944_2_2.wav 62400
+wavs_audio1526984421235_3_7.wav 39360
+wavs_audio1526984422523_2_7.wav 49920
+wavs_audio1526984426143_5_4.wav 48960
+wavs_audio1526984433191_2_3.wav 53760
+wavs_audio1526984435465_5_5.wav 56640
+wavs_audio1526984443372_5_6.wav 51840
+wavs_audio1526984444218_5_5.wav 47040
+wavs_audio1526984446328_3_8.wav 64320
+wavs_audio1526984455383_2_4.wav 51840
+wavs_audio1526984456002_3_7.wav 50880
+wavs_audio1526984456727_3_2.wav 34560
+wavs_audio1526984460345_5_6.wav 46080
+wavs_audio1526984469365_3_8.wav 53760
+wavs_audio1526984482150_3_3.wav 42240
+wavs_audio1526984485261_2_5.wav 45120
+wavs_audio1526984487880_4_2.wav 53760
+wavs_audio1526984492899_4_1.wav 41280
+wavs_audio1526984497103_5_7.wav 81600
+wavs_audio1526984500341_2_6.wav 54720
+wavs_audio1526984502455_6_1.wav 35520
+wavs_audio1526984504361_3_5.wav 44160
+wavs_audio1526984504722_4_4.wav 39360
+wavs_audio1526984506678_4_2.wav 58560
+wavs_audio1526984512402_6_2.wav 45120
+wavs_audio1526984513900_2_7.wav 48000
+wavs_audio1526984518655_4_1.wav 53760
+wavs_audio1526984523862_6_3.wav 44160
+wavs_audio1526984525492_4_3.wav 72000
+wavs_audio1526984528468_4_2.wav 49920
+wavs_audio1526984533901_6_4.wav 47040
+wavs_audio1526984535229_3_1.wav 44160
+wavs_audio1526984536286_4_3.wav 39360
+wavs_audio1526984538200_5_1.wav 64320
+wavs_audio1526984540680_4_4.wav 40320
+wavs_audio1526984546639_4_4.wav 40320
+wavs_audio1526984547590_5_2.wav 61440
+wavs_audio1526984552838_6_3.wav 64320
+wavs_audio1526984556510_4_5.wav 63360
+wavs_audio1526984559307_3_3.wav 43200
+wavs_audio1526984569143_5_1.wav 48000
+wavs_audio1526984570131_5_3.wav 60480
+wavs_audio1526984573839_6_4.wav 100800
+wavs_audio1526984575227_3_4.wav 52800
+wavs_audio1526984577387_5_1.wav 58560
+wavs_audio1526984579412_5_2.wav 54720
+wavs_audio1526984586922_5_3.wav 46080
+wavs_audio1526984588951_3_5.wav 60480
+wavs_audio1526984591785_5_2.wav 65280
+wavs_audio1526984595756_5_4.wav 42240

exp/asr_stats_raw_en_word/logdir/stats.10/train/stats_keys

@@ -0,0 +1,2 @@
+feats
+feats_lengths

exp/asr_stats_raw_en_word/logdir/stats.10/train/text_shape

@@ -0,0 +1,191 @@
+wavs_audio1526983626238_6_1.wav 21
+wavs_audio1526983636871_5_2.wav 32
+wavs_audio1526983641501_5_3.wav 29
+wavs_audio1526983643559_6_2.wav 23
+wavs_audio1526983652301_1_1.wav 18
+wavs_audio1526983653669_5_1.wav 27
+wavs_audio1526983663015_5_3.wav 29
+wavs_audio1526983668420_1_2.wav 15
+wavs_audio1526983671576_5_3.wav 29
+wavs_audio1526983674709_5_5.wav 27
+wavs_audio1526983682930_1_3.wav 9
+wavs_audio1526983683850_5_6.wav 27
+wavs_audio1526983684511_5_4.wav 29
+wavs_audio1526983687446_6_4.wav 35
+wavs_audio1526983694130_5_7.wav 29
+wavs_audio1526983695453_1_4.wav 20
+wavs_audio1526983711192_1_5.wav 17
+wavs_audio1526983718405_5_5.wav 27
+wavs_audio1526983722788_6_2.wav 23
+wavs_audio1526983726631_5_6.wav 27
+wavs_audio1526983730520_6_3.wav 33
+wavs_audio1526983732202_1_6.wav 11
+wavs_audio1526983741838_6_4.wav 35
+wavs_audio1526983744462_5_7.wav 29
+wavs_audio1526983759496_6_1.wav 21
+wavs_audio1526983771215_6_2.wav 23
+wavs_audio1526983778361_1_1.wav 18
+wavs_audio1526983780570_6_3.wav 33
+wavs_audio1526983788348_6_4.wav 35
+wavs_audio1526983790175_2_4.wav 21
+wavs_audio1526983800150_1_2.wav 15
+wavs_audio1526983802451_2_5.wav 14
+wavs_audio1526983812921_1_3.wav 9
+wavs_audio1526983815268_2_6.wav 20
+wavs_audio1526983824560_2_7.wav 20
+wavs_audio1526983825156_1_4.wav 20
+wavs_audio1526983850722_1_5.wav 17
+wavs_audio1526983866391_1_6.wav 11
+wavs_audio1526983882309_1_7.wav 31
+wavs_audio1526983886714_3_1.wav 15
+wavs_audio1526983895919_1_8.wav 33
+wavs_audio1526983898472_3_2.wav 13
+wavs_audio1526983911045_3_3.wav 15
+wavs_audio1526983930113_3_4.wav 13
+wavs_audio1526983931076_2_1.wav 23
+wavs_audio1526983945675_2_2.wav 23
+wavs_audio1526983972052_3_6.wav 20
+wavs_audio1526983981691_1_3.wav 9
+wavs_audio1526983990965_1_2.wav 15
+wavs_audio1526983992330_3_7.wav 13
+wavs_audio1526983996751_2_4.wav 21
+wavs_audio1526983997319_1_2.wav 15
+wavs_audio1526984003559_3_8.wav 13
+wavs_audio1526984007551_2_5.wav 14
+wavs_audio1526984011260_1_3.wav 9
+wavs_audio1526984017034_2_6.wav 20
+wavs_audio1526984027086_1_4.wav 20
+wavs_audio1526984029175_4_1.wav 18
+wavs_audio1526984031089_1_4.wav 20
+wavs_audio1526984031785_1_1.wav 18
+wavs_audio1526984036073_2_7.wav 20
+wavs_audio1526984040829_1_5.wav 17
+wavs_audio1526984044743_4_2.wav 20
+wavs_audio1526984055641_4_3.wav 18
+wavs_audio1526984071399_4_4.wav 12
+wavs_audio1526984082880_4_5.wav 20
+wavs_audio1526984083121_1_8.wav 33
+wavs_audio1526984083509_1_7.wav 31
+wavs_audio1526984106151_3_2.wav 13
+wavs_audio1526984114132_5_1.wav 27
+wavs_audio1526984121674_3_3.wav 15
+wavs_audio1526984123594_2_4.wav 21
+wavs_audio1526984135088_2_5.wav 14
+wavs_audio1526984139552_3_4.wav 13
+wavs_audio1526984144689_2_6.wav 20
+wavs_audio1526984147192_1_1.wav 18
+wavs_audio1526984148728_2_4.wav 21
+wavs_audio1526984153209_2_7.wav 20
+wavs_audio1526984156770_3_5.wav 20
+wavs_audio1526984169153_3_6.wav 20
+wavs_audio1526984171836_2_1.wav 23
+wavs_audio1526984173406_2_5.wav 14
+wavs_audio1526984173931_5_3.wav 29
+wavs_audio1526984179972_3_1.wav 15
+wavs_audio1526984186071_3_7.wav 13
+wavs_audio1526984186573_2_2.wav 23
+wavs_audio1526984197610_2_3.wav 21
+wavs_audio1526984200561_3_2.wav 13
+wavs_audio1526984206448_5_5.wav 27
+wavs_audio1526984214431_3_3.wav 15
+wavs_audio1526984217633_3_8.wav 13
+wavs_audio1526984223567_1_1.wav 18
+wavs_audio1526984236297_1_2.wav 15
+wavs_audio1526984237031_3_4.wav 13
+wavs_audio1526984241114_5_7.wav 29
+wavs_audio1526984243264_1_4.wav 20
+wavs_audio1526984245216_2_2.wav 23
+wavs_audio1526984248224_1_3.wav 9
+wavs_audio1526984249624_3_5.wav 20
+wavs_audio1526984255903_2_5.wav 14
+wavs_audio1526984256425_3_6.wav 20
+wavs_audio1526984262616_2_2.wav 23
+wavs_audio1526984265407_1_4.wav 20
+wavs_audio1526984281593_6_1.wav 21
+wavs_audio1526984282940_1_6.wav 11
+wavs_audio1526984295750_2_7.wav 20
+wavs_audio1526984296338_4_1.wav 18
+wavs_audio1526984301655_1_6.wav 11
+wavs_audio1526984307908_1_7.wav 31
+wavs_audio1526984311940_3_2.wav 13
+wavs_audio1526984318859_4_2.wav 20
+wavs_audio1526984321249_3_1.wav 15
+wavs_audio1526984323422_5_1.wav 27
+wavs_audio1526984323982_4_1.wav 18
+wavs_audio1526984326235_1_7.wav 31
+wavs_audio1526984327222_4_3.wav 18
+wavs_audio1526984333916_4_4.wav 12
+wavs_audio1526984334159_4_2.wav 20
+wavs_audio1526984337198_1_8.wav 33
+wavs_audio1526984337797_6_3.wav 33
+wavs_audio1526984345066_4_5.wav 20
+wavs_audio1526984345363_4_3.wav 18
+wavs_audio1526984349176_6_3.wav 33
+wavs_audio1526984351152_3_3.wav 15
+wavs_audio1526984354681_4_4.wav 12
+wavs_audio1526984354872_2_4.wav 21
+wavs_audio1526984357564_6_4.wav 35
+wavs_audio1526984363716_4_5.wav 20
+wavs_audio1526984365423_2_1.wav 23
+wavs_audio1526984372167_3_4.wav 13
+wavs_audio1526984375689_2_2.wav 23
+wavs_audio1526984377956_2_1.wav 23
+wavs_audio1526984381963_5_1.wav 27
+wavs_audio1526984384912_3_5.wav 20
+wavs_audio1526984392258_2_3.wav 21
+wavs_audio1526984400068_6_4.wav 35
+wavs_audio1526984402818_5_2.wav 32
+wavs_audio1526984404232_5_2.wav 32
+wavs_audio1526984407407_3_6.wav 20
+wavs_audio1526984414330_2_6.wav 20
+wavs_audio1526984416571_5_3.wav 29
+wavs_audio1526984417944_2_2.wav 23
+wavs_audio1526984421235_3_7.wav 13
+wavs_audio1526984422523_2_7.wav 20
+wavs_audio1526984426143_5_4.wav 29
+wavs_audio1526984433191_2_3.wav 21
+wavs_audio1526984435465_5_5.wav 27
+wavs_audio1526984443372_5_6.wav 27
+wavs_audio1526984444218_5_5.wav 27
+wavs_audio1526984446328_3_8.wav 13
+wavs_audio1526984455383_2_4.wav 21
+wavs_audio1526984456002_3_7.wav 13
+wavs_audio1526984456727_3_2.wav 13
+wavs_audio1526984460345_5_6.wav 27
+wavs_audio1526984469365_3_8.wav 13
+wavs_audio1526984482150_3_3.wav 15
+wavs_audio1526984485261_2_5.wav 14
+wavs_audio1526984487880_4_2.wav 20
+wavs_audio1526984492899_4_1.wav 18
+wavs_audio1526984497103_5_7.wav 29
+wavs_audio1526984500341_2_6.wav 20
+wavs_audio1526984502455_6_1.wav 21
+wavs_audio1526984504361_3_5.wav 20
+wavs_audio1526984504722_4_4.wav 12
+wavs_audio1526984506678_4_2.wav 20
+wavs_audio1526984512402_6_2.wav 23
+wavs_audio1526984513900_2_7.wav 20
+wavs_audio1526984518655_4_1.wav 18
+wavs_audio1526984523862_6_3.wav 33
+wavs_audio1526984525492_4_3.wav 18
+wavs_audio1526984528468_4_2.wav 20
+wavs_audio1526984533901_6_4.wav 35
+wavs_audio1526984535229_3_1.wav 15
+wavs_audio1526984536286_4_3.wav 18
+wavs_audio1526984538200_5_1.wav 27
+wavs_audio1526984540680_4_4.wav 12
+wavs_audio1526984546639_4_4.wav 12
+wavs_audio1526984547590_5_2.wav 32
+wavs_audio1526984552838_6_3.wav 33
+wavs_audio1526984556510_4_5.wav 20
+wavs_audio1526984559307_3_3.wav 15
+wavs_audio1526984569143_5_1.wav 27
+wavs_audio1526984570131_5_3.wav 29
+wavs_audio1526984573839_6_4.wav 35
+wavs_audio1526984575227_3_4.wav 13
+wavs_audio1526984577387_5_1.wav 27
+wavs_audio1526984579412_5_2.wav 32
+wavs_audio1526984586922_5_3.wav 29
+wavs_audio1526984588951_3_5.wav 20
+wavs_audio1526984591785_5_2.wav 32
+wavs_audio1526984595756_5_4.wav 29

exp/asr_stats_raw_en_word/logdir/stats.10/valid/batch_keys

@@ -0,0 +1,2 @@
+speech
+text

exp/asr_stats_raw_en_word/logdir/stats.10/valid/speech_shape

@@ -0,0 +1,24 @@
+wavs_audio1526983286623_5_7.wav 43200
+wavs_audio1526983324951_5_3.wav 90240
+wavs_audio1526983385142_4_5.wav 48000
+wavs_audio1526983478995_4_3.wav 50880
+wavs_audio1526983516946_5_1.wav 60480
+wavs_audio1526983544466_5_3.wav 67200
+wavs_audio1526983563031_3_8.wav 36480
+wavs_audio1526983564943_3_6.wav 41280
+wavs_audio1526983592590_4_2.wav 40320
+wavs_audio1526983663160_6_3.wav 60480
+wavs_audio1526983920025_1_1.wav 80640
+wavs_audio1526984094471_3_1.wav 41280
+wavs_audio1526984098387_1_7.wav 77760
+wavs_audio1526984102607_1_8.wav 72000
+wavs_audio1526984124218_1_8.wav 76800
+wavs_audio1526984189062_1_5.wav 79680
+wavs_audio1526984191576_5_4.wav 71040
+wavs_audio1526984208550_1_5.wav 53760
+wavs_audio1526984218128_2_7.wav 59520
+wavs_audio1526984225517_5_6.wav 73920
+wavs_audio1526984236438_2_5.wav 63360
+wavs_audio1526984273150_2_6.wav 57600
+wavs_audio1526984301196_6_2.wav 69120
+wavs_audio1526984345604_3_2.wav 45120

exp/asr_stats_raw_en_word/logdir/stats.10/valid/stats_keys

@@ -0,0 +1,2 @@
+feats
+feats_lengths

exp/asr_stats_raw_en_word/logdir/stats.10/valid/text_shape

@@ -0,0 +1,24 @@
+wavs_audio1526983286623_5_7.wav 29
+wavs_audio1526983324951_5_3.wav 29
+wavs_audio1526983385142_4_5.wav 20
+wavs_audio1526983478995_4_3.wav 18
+wavs_audio1526983516946_5_1.wav 27
+wavs_audio1526983544466_5_3.wav 29
+wavs_audio1526983563031_3_8.wav 13
+wavs_audio1526983564943_3_6.wav 20
+wavs_audio1526983592590_4_2.wav 20
+wavs_audio1526983663160_6_3.wav 33
+wavs_audio1526983920025_1_1.wav 18
+wavs_audio1526984094471_3_1.wav 15
+wavs_audio1526984098387_1_7.wav 31
+wavs_audio1526984102607_1_8.wav 33
+wavs_audio1526984124218_1_8.wav 33
+wavs_audio1526984189062_1_5.wav 17
+wavs_audio1526984191576_5_4.wav 29
+wavs_audio1526984208550_1_5.wav 17
+wavs_audio1526984218128_2_7.wav 20
+wavs_audio1526984225517_5_6.wav 27
+wavs_audio1526984236438_2_5.wav 14
+wavs_audio1526984273150_2_6.wav 20
+wavs_audio1526984301196_6_2.wav 23
+wavs_audio1526984345604_3_2.wav 13

exp/asr_stats_raw_en_word/logdir/stats.11.log

@@ -0,0 +1,495 @@
+# Running on r099.ib.bridges2.psc.edu
+# Started at Tue Dec 21 22:24:57 EST 2021
+# SLURMD_NODENAME=r099
+# SLURM_ARRAY_JOB_ID=5730432
+# SLURM_ARRAY_TASK_COUNT=32
+# SLURM_ARRAY_TASK_ID=11
+# SLURM_ARRAY_TASK_MAX=32
+# SLURM_ARRAY_TASK_MIN=1
+# SLURM_ARRAY_TASK_STEP=1
+# SLURM_CLUSTER_NAME=bridges2
+# SLURM_CONF=/var/spool/slurm/d/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=1
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GTIDS=0
+# SLURM_JOBID=5730464
+# SLURM_JOB_ACCOUNT=cis210027p
+# SLURM_JOB_CPUS_PER_NODE=1
+# SLURM_JOB_GID=24886
+# SLURM_JOB_ID=5730464
+# SLURM_JOB_NAME=stats.sh
+# SLURM_JOB_NODELIST=r099
+# SLURM_JOB_NUM_NODES=1
+# SLURM_JOB_PARTITION=RM-shared
+# SLURM_JOB_QOS=rm
+# SLURM_JOB_UID=82326
+# SLURM_JOB_USER=ganesank
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_CPU=2000
+# SLURM_NNODES=1
+# SLURM_NODEID=0
+# SLURM_NODELIST=r099
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/ocean/projects/cis210027p/ganesank/karthik_new/espnet/egs2/sinhala/asr1
+# SLURM_SUBMIT_HOST=br012.ib.bridges2.psc.edu
+# SLURM_TASKS_PER_NODE=1
+# SLURM_TASK_PID=6420
+# SLURM_TOPOLOGY_ADDR=r099
+# SLURM_TOPOLOGY_ADDR_PATTERN=node
+# SLURM_WORKING_CLUSTER=bridges2:br003:6814:9216:109
+# python3 -m espnet2.bin.asr_train --collect_stats true --use_preprocessor true --bpemodel none --token_type word --token_list data/en_token_list/word/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --train_data_path_and_name_and_type dump/raw/train/wav.scp,speech,sound --train_data_path_and_name_and_type dump/raw/train/text,text,text --valid_data_path_and_name_and_type dump/raw/valid/wav.scp,speech,sound --valid_data_path_and_name_and_type dump/raw/valid/text,text,text --train_shape_file exp/asr_stats_raw_en_word/logdir/train.11.scp --valid_shape_file exp/asr_stats_raw_en_word/logdir/valid.11.scp --output_dir exp/asr_stats_raw_en_word/logdir/stats.11 --config conf/train_asr.yaml --frontend_conf fs=16k 
+/ocean/projects/cis210027p/ganesank/karthik_new/espnet/tools/venv/bin/python3 /ocean/projects/cis210027p/ganesank/karthik_new/espnet/espnet2/bin/asr_train.py --collect_stats true --use_preprocessor true --bpemodel none --token_type word --token_list data/en_token_list/word/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --train_data_path_and_name_and_type dump/raw/train/wav.scp,speech,sound --train_data_path_and_name_and_type dump/raw/train/text,text,text --valid_data_path_and_name_and_type dump/raw/valid/wav.scp,speech,sound --valid_data_path_and_name_and_type dump/raw/valid/text,text,text --train_shape_file exp/asr_stats_raw_en_word/logdir/train.11.scp --valid_shape_file exp/asr_stats_raw_en_word/logdir/valid.11.scp --output_dir exp/asr_stats_raw_en_word/logdir/stats.11 --config conf/train_asr.yaml --frontend_conf fs=16k
+[r099] 2021-12-21 22:25:08,042 (asr:382) INFO: Vocabulary size: 40
+[r099] 2021-12-21 22:25:09,019 (abs_task:1132) INFO: pytorch.version=1.8.1+cu102, cuda.available=False, cudnn.version=7605, cudnn.benchmark=False, cudnn.deterministic=True
+[r099] 2021-12-21 22:25:09,024 (abs_task:1133) INFO: Model structure:
+ESPnetASRModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=512, hop_length=128, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (time_warp): TimeWarp(window=5, mode=bicubic)
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 30], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxis(mask_width_range=[0, 40], num_mask=2, axis=time)
+  )
+  (normalize): UtteranceMVN(norm_means=True, norm_vars=False)
+  (encoder): TransformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 256, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=4864, out_features=256, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (9): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (10): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (11): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(40, 256)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=256, out_features=40, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=256, out_features=40, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetASRModel
+    Total Number of model parameters: 27.12 M
+    Number of trainable parameters: 27.12 M (100.0%)
+    Size: 108.49 MB
+    Type: torch.float32
+[r099] 2021-12-21 22:25:09,024 (abs_task:1136) INFO: Optimizer:
+Adam (
+Parameter Group 0
+    amsgrad: False
+    betas: (0.9, 0.999)
+    eps: 1e-08
+    initial_lr: 0.0002
+    lr: 8e-09
+    weight_decay: 0
+)
+[r099] 2021-12-21 22:25:09,024 (abs_task:1137) INFO: Scheduler: WarmupLR(warmup_steps=25000)
+[r099] 2021-12-21 22:25:09,027 (abs_task:1146) INFO: Saving the configuration in exp/asr_stats_raw_en_word/logdir/stats.11/config.yaml
+[r099] 2021-12-21 22:25:09,037 (abs_task:1157) INFO: Namespace(config='conf/train_asr.yaml', print_config=False, log_level='INFO', dry_run=False, iterator_type='sequence', output_dir='exp/asr_stats_raw_en_word/logdir/stats.11', ngpu=0, seed=0, num_workers=1, num_att_plot=3, dist_backend='nccl', dist_init_method='env://', dist_world_size=None, dist_rank=None, local_rank=None, dist_master_addr=None, dist_master_port=None, dist_launcher=None, multiprocessing_distributed=False, unused_parameters=False, sharded_ddp=False, cudnn_enabled=True, cudnn_benchmark=False, cudnn_deterministic=True, collect_stats=True, write_collected_feats=False, max_epoch=50, patience=None, val_scheduler_criterion=('valid', 'loss'), early_stopping_criterion=('valid', 'loss', 'min'), best_model_criterion=[('train', 'loss', 'min'), ('valid', 'loss', 'min'), ('train', 'acc', 'max'), ('valid', 'acc', 'max')], keep_nbest_models=5, grad_clip=5.0, grad_clip_type=2.0, grad_noise=False, accum_grad=1, no_forward_run=False, resume=False, train_dtype='float32', use_amp=False, log_interval=None, use_tensorboard=True, use_wandb=False, wandb_project=None, wandb_id=None, wandb_entity=None, wandb_name=None, wandb_model_log_interval=-1, detect_anomaly=False, pretrain_path=None, init_param=[], ignore_init_mismatch=False, freeze_param=[], num_iters_per_epoch=None, batch_size=20, valid_batch_size=None, batch_bins=1000000, valid_batch_bins=None, train_shape_file=['exp/asr_stats_raw_en_word/logdir/train.11.scp'], valid_shape_file=['exp/asr_stats_raw_en_word/logdir/valid.11.scp'], batch_type='folded', valid_batch_type=None, fold_length=[], sort_in_batch='descending', sort_batch='descending', multiple_iterator=False, chunk_length=500, chunk_shift_ratio=0.5, num_cache_chunks=1024, train_data_path_and_name_and_type=[('dump/raw/train/wav.scp', 'speech', 'sound'), ('dump/raw/train/text', 'text', 'text')], valid_data_path_and_name_and_type=[('dump/raw/valid/wav.scp', 'speech', 'sound'), ('dump/raw/valid/text', 'text', 'text')], allow_variable_data_keys=False, max_cache_size=0.0, max_cache_fd=32, valid_max_cache_size=None, optim='adam', optim_conf={'lr': 0.0002}, scheduler='warmuplr', scheduler_conf={'warmup_steps': 25000}, token_list=['<blank>', '<unk>', '්', 'න', 'ම', 'ක', 'ල', 'ි', 'ු', 'ග', 'ේ', 'ර', 'ත', 'ද', 'ව', 'ට', 'ඕ', 'ී', 'ප', 'ය', 'ෙ', 'ස', 'ණ', 'ා', 'ැ', 'RequestAcc.balance', 'Moneywithdraw', 'Moneydeposit', 'Moneytransfer', 'Billpayments', 'බ', 'ඉ', 'ශ', 'ෂ', 'ඩ', 'Creditcardpayments', 'එ', '\u200d', 'හ', '<sos/eos>'], init=None, input_size=None, ctc_conf={'dropout_rate': 0.0, 'ctc_type': 'builtin', 'reduce': True, 'ignore_nan_grad': True}, model_conf={'ctc_weight': 0.5, 'ignore_id': -1, 'lsm_weight': 0.0, 'length_normalized_loss': False, 'report_cer': True, 'report_wer': True, 'sym_space': '<space>', 'sym_blank': '<blank>', 'extract_feats_in_collect_stats': True}, use_preprocessor=True, token_type='word', bpemodel=None, non_linguistic_symbols=None, cleaner=None, g2p=None, speech_volume_normalize=None, rir_scp=None, rir_apply_prob=1.0, noise_scp=None, noise_apply_prob=1.0, noise_db_range='13_15', frontend='default', frontend_conf={'fs': '16k'}, specaug='specaug', specaug_conf={'apply_time_warp': True, 'time_warp_window': 5, 'time_warp_mode': 'bicubic', 'apply_freq_mask': True, 'freq_mask_width_range': [0, 30], 'num_freq_mask': 2, 'apply_time_mask': True, 'time_mask_width_range': [0, 40], 'num_time_mask': 2}, normalize='utterance_mvn', normalize_conf={}, preencoder=None, preencoder_conf={}, encoder='transformer', encoder_conf={'output_size': 256, 'attention_heads': 4, 'linear_units': 2048, 'num_blocks': 12, 'dropout_rate': 0.1, 'positional_dropout_rate': 0.1, 'attention_dropout_rate': 0.0, 'input_layer': 'conv2d', 'normalize_before': True}, postencoder=None, postencoder_conf={}, decoder='transformer', decoder_conf={'attention_heads': 4, 'linear_units': 2048, 'num_blocks': 6, 'dropout_rate': 0.1, 'positional_dropout_rate': 0.1, 'self_attention_dropout_rate': 0.0, 'src_attention_dropout_rate': 0.0}, required=['output_dir', 'token_list'], version='0.10.3a3', distributed=False)
+# Accounting: begin_time=1640143497
+# Accounting: end_time=1640143519
+# Accounting: time=22 threads=1
+# Finished at Tue Dec 21 22:25:19 EST 2021 with status 0

exp/asr_stats_raw_en_word/logdir/stats.11/config.yaml

@@ -0,0 +1,231 @@
+config: conf/train_asr.yaml
+print_config: false
+log_level: INFO
+dry_run: false
+iterator_type: sequence
+output_dir: exp/asr_stats_raw_en_word/logdir/stats.11
+ngpu: 0
+seed: 0
+num_workers: 1
+num_att_plot: 3
+dist_backend: nccl
+dist_init_method: env://
+dist_world_size: null
+dist_rank: null
+local_rank: null
+dist_master_addr: null
+dist_master_port: null
+dist_launcher: null
+multiprocessing_distributed: false
+unused_parameters: false
+sharded_ddp: false
+cudnn_enabled: true
+cudnn_benchmark: false
+cudnn_deterministic: true
+collect_stats: true
+write_collected_feats: false
+max_epoch: 50
+patience: null
+val_scheduler_criterion:
+- valid
+- loss
+early_stopping_criterion:
+- valid
+- loss
+- min
+best_model_criterion:
+-   - train
+    - loss
+    - min
+-   - valid
+    - loss
+    - min
+-   - train
+    - acc
+    - max
+-   - valid
+    - acc
+    - max
+keep_nbest_models: 5
+grad_clip: 5.0
+grad_clip_type: 2.0
+grad_noise: false
+accum_grad: 1
+no_forward_run: false
+resume: false
+train_dtype: float32
+use_amp: false
+log_interval: null
+use_tensorboard: true
+use_wandb: false
+wandb_project: null
+wandb_id: null
+wandb_entity: null
+wandb_name: null
+wandb_model_log_interval: -1
+detect_anomaly: false
+pretrain_path: null
+init_param: []
+ignore_init_mismatch: false
+freeze_param: []
+num_iters_per_epoch: null
+batch_size: 20
+valid_batch_size: null
+batch_bins: 1000000
+valid_batch_bins: null
+train_shape_file:
+- exp/asr_stats_raw_en_word/logdir/train.11.scp
+valid_shape_file:
+- exp/asr_stats_raw_en_word/logdir/valid.11.scp
+batch_type: folded
+valid_batch_type: null
+fold_length: []
+sort_in_batch: descending
+sort_batch: descending
+multiple_iterator: false
+chunk_length: 500
+chunk_shift_ratio: 0.5
+num_cache_chunks: 1024
+train_data_path_and_name_and_type:
+-   - dump/raw/train/wav.scp
+    - speech
+    - sound
+-   - dump/raw/train/text
+    - text
+    - text
+valid_data_path_and_name_and_type:
+-   - dump/raw/valid/wav.scp
+    - speech
+    - sound
+-   - dump/raw/valid/text
+    - text
+    - text
+allow_variable_data_keys: false
+max_cache_size: 0.0
+max_cache_fd: 32
+valid_max_cache_size: null
+optim: adam
+optim_conf:
+    lr: 0.0002
+scheduler: warmuplr
+scheduler_conf:
+    warmup_steps: 25000
+token_list:
+- <blank>
+- <unk>
+- ්
+- න
+- ම
+- ක
+- ල
+- ි
+- ු
+- ග
+- ේ
+- ර
+- ත
+- ද
+- ව
+- ට
+- ඕ
+- ී
+- ප
+- ය
+- ෙ
+- ස
+- ණ
+- ා
+- ැ
+- RequestAcc.balance
+- Moneywithdraw
+- Moneydeposit
+- Moneytransfer
+- Billpayments
+- බ
+- ඉ
+- ශ
+- ෂ
+- ඩ
+- Creditcardpayments
+- එ
+- ‍
+- හ
+- <sos/eos>
+init: null
+input_size: null
+ctc_conf:
+    dropout_rate: 0.0
+    ctc_type: builtin
+    reduce: true
+    ignore_nan_grad: true
+model_conf:
+    ctc_weight: 0.5
+    ignore_id: -1
+    lsm_weight: 0.0
+    length_normalized_loss: false
+    report_cer: true
+    report_wer: true
+    sym_space: <space>
+    sym_blank: <blank>
+    extract_feats_in_collect_stats: true
+use_preprocessor: true
+token_type: word
+bpemodel: null
+non_linguistic_symbols: null
+cleaner: null
+g2p: null
+speech_volume_normalize: null
+rir_scp: null
+rir_apply_prob: 1.0
+noise_scp: null
+noise_apply_prob: 1.0
+noise_db_range: '13_15'
+frontend: default
+frontend_conf:
+    fs: 16k
+specaug: specaug
+specaug_conf:
+    apply_time_warp: true
+    time_warp_window: 5
+    time_warp_mode: bicubic
+    apply_freq_mask: true
+    freq_mask_width_range:
+    - 0
+    - 30
+    num_freq_mask: 2
+    apply_time_mask: true
+    time_mask_width_range:
+    - 0
+    - 40
+    num_time_mask: 2
+normalize: utterance_mvn
+normalize_conf: {}
+preencoder: null
+preencoder_conf: {}
+encoder: transformer
+encoder_conf:
+    output_size: 256
+    attention_heads: 4
+    linear_units: 2048
+    num_blocks: 12
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    attention_dropout_rate: 0.0
+    input_layer: conv2d
+    normalize_before: true
+postencoder: null
+postencoder_conf: {}
+decoder: transformer
+decoder_conf:
+    attention_heads: 4
+    linear_units: 2048
+    num_blocks: 6
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    self_attention_dropout_rate: 0.0
+    src_attention_dropout_rate: 0.0
+required:
+- output_dir
+- token_list
+version: 0.10.3a3
+distributed: false

exp/asr_stats_raw_en_word/logdir/stats.11/train/batch_keys

@@ -0,0 +1,2 @@
+speech
+text

exp/asr_stats_raw_en_word/logdir/stats.11/train/speech_shape

@@ -0,0 +1,191 @@
+wavs_audio1526984602629_3_6.wav 51840
+wavs_audio1526984606896_5_5.wav 61440
+wavs_audio1526984610105_5_3.wav 57600
+wavs_audio1526984615099_3_7.wav 38400
+wavs_audio1526984621281_5_6.wav 77760
+wavs_audio1526984629715_5_7.wav 48960
+wavs_audio1526984636403_5_7.wav 62400
+wavs_audio1526984643040_6_1.wav 46080
+wavs_audio1526984650962_6_2.wav 47040
+wavs_audio1526984652622_4_1.wav 46080
+wavs_audio1526984654940_1_1.wav 107520
+wavs_audio1526984655280_5_5.wav 60480
+wavs_audio1526984661490_6_3.wav 50880
+wavs_audio1526984663485_6_1.wav 82560
+wavs_audio1526984670577_5_6.wav 65280
+wavs_audio1526984676449_6_4.wav 51840
+wavs_audio1526984685159_5_7.wav 59520
+wavs_audio1526984715970_1_3.wav 70080
+wavs_audio1526984729286_6_2.wav 59520
+wavs_audio1526984733224_6_1.wav 98880
+wavs_audio1526984737083_1_4.wav 71040
+wavs_audio1526984743061_6_3.wav 58560
+wavs_audio1526984747137_1_1.wav 81600
+wavs_audio1526984750940_4_1.wav 31680
+wavs_audio1526984753531_1_5.wav 65280
+wavs_audio1526984759961_6_4.wav 75840
+wavs_audio1526984760580_6_3.wav 59520
+wavs_audio1526984767669_1_2.wav 74880
+wavs_audio1526984769766_1_6.wav 49920
+wavs_audio1526984772542_4_2.wav 88320
+wavs_audio1526984782069_1_3.wav 40320
+wavs_audio1526984788067_4_3.wav 52800
+wavs_audio1526984796332_1_4.wav 53760
+wavs_audio1526984812263_4_5.wav 38400
+wavs_audio1526984827811_2_1.wav 70080
+wavs_audio1526984836904_5_1.wav 76800
+wavs_audio1526984851904_5_2.wav 66240
+wavs_audio1526984854445_2_3.wav 78720
+wavs_audio1526984863287_1_7.wav 69120
+wavs_audio1526984867484_2_4.wav 83520
+wavs_audio1526984874070_5_3.wav 54720
+wavs_audio1526984876105_1_8.wav 66240
+wavs_audio1526984884127_2_5.wav 63360
+wavs_audio1526984889424_5_4.wav 56640
+wavs_audio1526984897049_2_6.wav 67200
+wavs_audio1526984918482_2_7.wav 76800
+wavs_audio1526984937456_2_3.wav 40320
+wavs_audio1526984940663_5_4.wav 37440
+wavs_audio1526984953188_1_1.wav 60480
+wavs_audio1526984954166_2_2.wav 56640
+wavs_audio1526984955453_1_1.wav 65280
+wavs_audio1526984966116_2_3.wav 56640
+wavs_audio1526984966716_1_2.wav 35520
+wavs_audio1526984979832_1_3.wav 33600
+wavs_audio1526984984486_3_2.wav 63360
+wavs_audio1526984987570_2_5.wav 63360
+wavs_audio1526984989387_1_3.wav 50880
+wavs_audio1526984997187_5_5.wav 61440
+wavs_audio1526985010117_2_6.wav 46080
+wavs_audio1526985013186_5_6.wav 54720
+wavs_audio1526985016193_1_5.wav 36480
+wavs_audio1526985021031_2_7.wav 51840
+wavs_audio1526985022105_1_5.wav 61440
+wavs_audio1526985026556_3_5.wav 72000
+wavs_audio1526985026940_1_6.wav 27840
+wavs_audio1526985027741_5_7.wav 52800
+wavs_audio1526985034297_1_6.wav 45120
+wavs_audio1526985037569_1_7.wav 40320
+wavs_audio1526985038725_3_6.wav 73920
+wavs_audio1526985048499_1_7.wav 69120
+wavs_audio1526985051021_1_8.wav 52800
+wavs_audio1526985056080_3_2.wav 47040
+wavs_audio1526985062105_1_8.wav 67200
+wavs_audio1526985063328_2_4.wav 36480
+wavs_audio1526985066949_3_3.wav 48960
+wavs_audio1526985075041_3_8.wav 73920
+wavs_audio1526985076424_6_1.wav 64320
+wavs_audio1526985078460_2_4.wav 57600
+wavs_audio1526985099760_4_1.wav 75840
+wavs_audio1526985114121_3_6.wav 50880
+wavs_audio1526985114292_4_2.wav 72000
+wavs_audio1526985125749_3_7.wav 41280
+wavs_audio1526985125890_4_3.wav 78720
+wavs_audio1526985133561_3_1.wav 102720
+wavs_audio1526985139204_3_8.wav 38400
+wavs_audio1526985140297_2_6.wav 48960
+wavs_audio1526985145800_4_4.wav 51840
+wavs_audio1526985154609_1_2.wav 55680
+wavs_audio1526985154890_3_3.wav 59520
+wavs_audio1526985160301_6_2.wav 42240
+wavs_audio1526985163521_3_4.wav 27840
+wavs_audio1526985165400_2_7.wav 57600
+wavs_audio1526985168489_1_3.wav 26880
+wavs_audio1526985181875_3_5.wav 31680
+wavs_audio1526985182786_6_3.wav 59520
+wavs_audio1526985191547_3_6.wav 31680
+wavs_audio1526985192554_4_4.wav 48000
+wavs_audio1526985192555_4_4.wav 48000
+wavs_audio1526985198227_6_4.wav 56640
+wavs_audio1526985199398_3_7.wav 29760
+wavs_audio1526985200453_5_1.wav 86400
+wavs_audio1526985203912_4_5.wav 48960
+wavs_audio1526985208375_3_8.wav 24960
+wavs_audio1526985212377_1_5.wav 31680
+wavs_audio1526985219145_3_3.wav 49920
+wavs_audio1526985221101_1_6.wav 48960
+wavs_audio1526985223090_5_2.wav 83520
+wavs_audio1526985229351_4_2.wav 39360
+wavs_audio1526985237111_3_4.wav 54720
+wavs_audio1526985239172_5_3.wav 79680
+wavs_audio1526985248309_5_2.wav 69120
+wavs_audio1526985249863_3_5.wav 51840
+wavs_audio1526985253528_1_8.wav 51840
+wavs_audio1526985254519_4_3.wav 27840
+wavs_audio1526985273327_5_5.wav 83520
+wavs_audio1526985279112_5_4.wav 72000
+wavs_audio1526985281186_3_7.wav 44160
+wavs_audio1526985288394_4_5.wav 33600
+wavs_audio1526985289109_2_1.wav 41280
+wavs_audio1526985291276_3_8.wav 46080
+wavs_audio1526985293509_5_5.wav 61440
+wavs_audio1526985301946_2_2.wav 56640
+wavs_audio1526985308192_1_5.wav 63360
+wavs_audio1526985308781_1_2.wav 69120
+wavs_audio1526985312760_2_3.wav 48960
+wavs_audio1526985315798_1_3.wav 56640
+wavs_audio1526985321737_1_6.wav 50880
+wavs_audio1526985321887_2_4.wav 53760
+wavs_audio1526985326172_1_3.wav 51840
+wavs_audio1526985333614_1_7.wav 75840
+wavs_audio1526985334281_4_3.wav 48960
+wavs_audio1526985336955_5_7.wav 67200
+wavs_audio1526985347270_1_8.wav 74880
+wavs_audio1526985352099_2_7.wav 49920
+wavs_audio1526985363089_1_6.wav 51840
+wavs_audio1526985369835_6_1.wav 72000
+wavs_audio1526985370370_5_4.wav 42240
+wavs_audio1526985372359_1_1.wav 63360
+wavs_audio1526985373862_3_1.wav 84480
+wavs_audio1526985386612_5_5.wav 46080
+wavs_audio1526985387973_6_2.wav 89280
+wavs_audio1526985394561_5_1.wav 58560
+wavs_audio1526985396204_5_6.wav 45120
+wavs_audio1526985396337_3_6.wav 37440
+wavs_audio1526985404906_1_8.wav 61440
+wavs_audio1526985412257_2_1.wav 46080
+wavs_audio1526985415661_5_7.wav 98880
+wavs_audio1526985416723_6_4.wav 80640
+wavs_audio1526985425547_5_6.wav 72960
+wavs_audio1526985436519_5_7.wav 64320
+wavs_audio1526985437223_6_1.wav 43200
+wavs_audio1526985443344_2_5.wav 44160
+wavs_audio1526985447143_4_3.wav 36480
+wavs_audio1526985449392_6_2.wav 57600
+wavs_audio1526985452841_6_1.wav 58560
+wavs_audio1526985453486_2_6.wav 46080
+wavs_audio1526985455235_4_4.wav 29760
+wavs_audio1526985457216_1_5.wav 66240
+wavs_audio1526985462205_4_5.wav 41280
+wavs_audio1526985464114_2_7.wav 50880
+wavs_audio1526985469912_1_4.wav 61440
+wavs_audio1526985473990_6_3.wav 65280
+wavs_audio1526985477187_6_4.wav 49920
+wavs_audio1526985482607_1_2.wav 38400
+wavs_audio1526985482769_5_5.wav 56640
+wavs_audio1526985484101_1_5.wav 53760
+wavs_audio1526985492075_1_3.wav 28800
+wavs_audio1526985492938_5_6.wav 58560
+wavs_audio1526985500156_1_4.wav 44160
+wavs_audio1526985504537_5_7.wav 55680
+wavs_audio1526985507902_1_7.wav 66240
+wavs_audio1526985509955_1_5.wav 37440
+wavs_audio1526985511640_1_7.wav 70080
+wavs_audio1526985516630_1_6.wav 24000
+wavs_audio1526985525691_6_1.wav 47040
+wavs_audio1526985526757_1_8.wav 67200
+wavs_audio1526985538442_6_2.wav 53760
+wavs_audio1526985539179_6_1.wav 46080
+wavs_audio1526985546134_2_4.wav 47040
+wavs_audio1526985548388_4_1.wav 48960
+wavs_audio1526985549115_6_2.wav 55680
+wavs_audio1526985550464_1_8.wav 57600
+wavs_audio1526985552389_1_1.wav 98880
+wavs_audio1526985555793_2_5.wav 52800
+wavs_audio1526985564579_2_6.wav 40320
+wavs_audio1526985568317_6_4.wav 65280
+wavs_audio1526985574403_2_7.wav 46080
+wavs_audio1526985577889_1_2.wav 93120
+wavs_audio1526985586019_4_3.wav 44160
+wavs_audio1526985604764_3_3.wav 37440
+wavs_audio1526985612198_4_4.wav 40320

exp/asr_stats_raw_en_word/logdir/stats.11/train/stats_keys

@@ -0,0 +1,2 @@
+feats
+feats_lengths

exp/asr_stats_raw_en_word/logdir/stats.11/train/text_shape

@@ -0,0 +1,191 @@
+wavs_audio1526984602629_3_6.wav 20
+wavs_audio1526984606896_5_5.wav 27
+wavs_audio1526984610105_5_3.wav 29
+wavs_audio1526984615099_3_7.wav 13
+wavs_audio1526984621281_5_6.wav 27
+wavs_audio1526984629715_5_7.wav 29
+wavs_audio1526984636403_5_7.wav 29
+wavs_audio1526984643040_6_1.wav 21
+wavs_audio1526984650962_6_2.wav 23
+wavs_audio1526984652622_4_1.wav 18
+wavs_audio1526984654940_1_1.wav 18
+wavs_audio1526984655280_5_5.wav 27
+wavs_audio1526984661490_6_3.wav 33
+wavs_audio1526984663485_6_1.wav 21
+wavs_audio1526984670577_5_6.wav 27
+wavs_audio1526984676449_6_4.wav 35
+wavs_audio1526984685159_5_7.wav 29
+wavs_audio1526984715970_1_3.wav 9
+wavs_audio1526984729286_6_2.wav 23
+wavs_audio1526984733224_6_1.wav 21
+wavs_audio1526984737083_1_4.wav 20
+wavs_audio1526984743061_6_3.wav 33
+wavs_audio1526984747137_1_1.wav 18
+wavs_audio1526984750940_4_1.wav 18
+wavs_audio1526984753531_1_5.wav 17
+wavs_audio1526984759961_6_4.wav 35
+wavs_audio1526984760580_6_3.wav 33
+wavs_audio1526984767669_1_2.wav 15
+wavs_audio1526984769766_1_6.wav 11
+wavs_audio1526984772542_4_2.wav 20
+wavs_audio1526984782069_1_3.wav 9
+wavs_audio1526984788067_4_3.wav 18
+wavs_audio1526984796332_1_4.wav 20
+wavs_audio1526984812263_4_5.wav 20
+wavs_audio1526984827811_2_1.wav 23
+wavs_audio1526984836904_5_1.wav 27
+wavs_audio1526984851904_5_2.wav 32
+wavs_audio1526984854445_2_3.wav 21
+wavs_audio1526984863287_1_7.wav 31
+wavs_audio1526984867484_2_4.wav 21
+wavs_audio1526984874070_5_3.wav 29
+wavs_audio1526984876105_1_8.wav 33
+wavs_audio1526984884127_2_5.wav 14
+wavs_audio1526984889424_5_4.wav 29
+wavs_audio1526984897049_2_6.wav 20
+wavs_audio1526984918482_2_7.wav 20
+wavs_audio1526984937456_2_3.wav 21
+wavs_audio1526984940663_5_4.wav 29
+wavs_audio1526984953188_1_1.wav 18
+wavs_audio1526984954166_2_2.wav 23
+wavs_audio1526984955453_1_1.wav 18
+wavs_audio1526984966116_2_3.wav 21
+wavs_audio1526984966716_1_2.wav 15
+wavs_audio1526984979832_1_3.wav 9
+wavs_audio1526984984486_3_2.wav 13
+wavs_audio1526984987570_2_5.wav 14
+wavs_audio1526984989387_1_3.wav 9
+wavs_audio1526984997187_5_5.wav 27
+wavs_audio1526985010117_2_6.wav 20
+wavs_audio1526985013186_5_6.wav 27
+wavs_audio1526985016193_1_5.wav 17
+wavs_audio1526985021031_2_7.wav 20
+wavs_audio1526985022105_1_5.wav 17
+wavs_audio1526985026556_3_5.wav 20
+wavs_audio1526985026940_1_6.wav 11
+wavs_audio1526985027741_5_7.wav 29
+wavs_audio1526985034297_1_6.wav 11
+wavs_audio1526985037569_1_7.wav 31
+wavs_audio1526985038725_3_6.wav 20
+wavs_audio1526985048499_1_7.wav 31
+wavs_audio1526985051021_1_8.wav 33
+wavs_audio1526985056080_3_2.wav 13
+wavs_audio1526985062105_1_8.wav 33
+wavs_audio1526985063328_2_4.wav 21
+wavs_audio1526985066949_3_3.wav 15
+wavs_audio1526985075041_3_8.wav 13
+wavs_audio1526985076424_6_1.wav 21
+wavs_audio1526985078460_2_4.wav 21
+wavs_audio1526985099760_4_1.wav 18
+wavs_audio1526985114121_3_6.wav 20
+wavs_audio1526985114292_4_2.wav 20
+wavs_audio1526985125749_3_7.wav 13
+wavs_audio1526985125890_4_3.wav 18
+wavs_audio1526985133561_3_1.wav 15
+wavs_audio1526985139204_3_8.wav 13
+wavs_audio1526985140297_2_6.wav 20
+wavs_audio1526985145800_4_4.wav 12
+wavs_audio1526985154609_1_2.wav 15
+wavs_audio1526985154890_3_3.wav 15
+wavs_audio1526985160301_6_2.wav 23
+wavs_audio1526985163521_3_4.wav 13
+wavs_audio1526985165400_2_7.wav 20
+wavs_audio1526985168489_1_3.wav 9
+wavs_audio1526985181875_3_5.wav 20
+wavs_audio1526985182786_6_3.wav 33
+wavs_audio1526985191547_3_6.wav 20
+wavs_audio1526985192554_4_4.wav 12
+wavs_audio1526985192555_4_4.wav 12
+wavs_audio1526985198227_6_4.wav 35
+wavs_audio1526985199398_3_7.wav 13
+wavs_audio1526985200453_5_1.wav 27
+wavs_audio1526985203912_4_5.wav 20
+wavs_audio1526985208375_3_8.wav 13
+wavs_audio1526985212377_1_5.wav 17
+wavs_audio1526985219145_3_3.wav 15
+wavs_audio1526985221101_1_6.wav 11
+wavs_audio1526985223090_5_2.wav 32
+wavs_audio1526985229351_4_2.wav 20
+wavs_audio1526985237111_3_4.wav 13
+wavs_audio1526985239172_5_3.wav 29
+wavs_audio1526985248309_5_2.wav 32
+wavs_audio1526985249863_3_5.wav 20
+wavs_audio1526985253528_1_8.wav 33
+wavs_audio1526985254519_4_3.wav 18
+wavs_audio1526985273327_5_5.wav 27
+wavs_audio1526985279112_5_4.wav 29
+wavs_audio1526985281186_3_7.wav 13
+wavs_audio1526985288394_4_5.wav 20
+wavs_audio1526985289109_2_1.wav 23
+wavs_audio1526985291276_3_8.wav 13
+wavs_audio1526985293509_5_5.wav 27
+wavs_audio1526985301946_2_2.wav 23
+wavs_audio1526985308192_1_5.wav 17
+wavs_audio1526985308781_1_2.wav 15
+wavs_audio1526985312760_2_3.wav 21
+wavs_audio1526985315798_1_3.wav 9
+wavs_audio1526985321737_1_6.wav 11
+wavs_audio1526985321887_2_4.wav 21
+wavs_audio1526985326172_1_3.wav 9
+wavs_audio1526985333614_1_7.wav 31
+wavs_audio1526985334281_4_3.wav 18
+wavs_audio1526985336955_5_7.wav 29
+wavs_audio1526985347270_1_8.wav 33
+wavs_audio1526985352099_2_7.wav 20
+wavs_audio1526985363089_1_6.wav 11
+wavs_audio1526985369835_6_1.wav 21
+wavs_audio1526985370370_5_4.wav 29
+wavs_audio1526985372359_1_1.wav 18
+wavs_audio1526985373862_3_1.wav 15
+wavs_audio1526985386612_5_5.wav 27
+wavs_audio1526985387973_6_2.wav 23
+wavs_audio1526985394561_5_1.wav 27
+wavs_audio1526985396204_5_6.wav 27
+wavs_audio1526985396337_3_6.wav 20
+wavs_audio1526985404906_1_8.wav 33
+wavs_audio1526985412257_2_1.wav 23
+wavs_audio1526985415661_5_7.wav 29
+wavs_audio1526985416723_6_4.wav 35
+wavs_audio1526985425547_5_6.wav 27
+wavs_audio1526985436519_5_7.wav 29
+wavs_audio1526985437223_6_1.wav 21
+wavs_audio1526985443344_2_5.wav 14
+wavs_audio1526985447143_4_3.wav 18
+wavs_audio1526985449392_6_2.wav 23
+wavs_audio1526985452841_6_1.wav 21
+wavs_audio1526985453486_2_6.wav 20
+wavs_audio1526985455235_4_4.wav 12
+wavs_audio1526985457216_1_5.wav 17
+wavs_audio1526985462205_4_5.wav 20
+wavs_audio1526985464114_2_7.wav 20
+wavs_audio1526985469912_1_4.wav 20
+wavs_audio1526985473990_6_3.wav 33
+wavs_audio1526985477187_6_4.wav 35
+wavs_audio1526985482607_1_2.wav 15
+wavs_audio1526985482769_5_5.wav 27
+wavs_audio1526985484101_1_5.wav 17
+wavs_audio1526985492075_1_3.wav 9
+wavs_audio1526985492938_5_6.wav 27
+wavs_audio1526985500156_1_4.wav 20
+wavs_audio1526985504537_5_7.wav 29
+wavs_audio1526985507902_1_7.wav 31
+wavs_audio1526985509955_1_5.wav 17
+wavs_audio1526985511640_1_7.wav 31
+wavs_audio1526985516630_1_6.wav 11
+wavs_audio1526985525691_6_1.wav 21
+wavs_audio1526985526757_1_8.wav 33
+wavs_audio1526985538442_6_2.wav 23
+wavs_audio1526985539179_6_1.wav 21
+wavs_audio1526985546134_2_4.wav 21
+wavs_audio1526985548388_4_1.wav 18
+wavs_audio1526985549115_6_2.wav 23
+wavs_audio1526985550464_1_8.wav 33
+wavs_audio1526985552389_1_1.wav 18
+wavs_audio1526985555793_2_5.wav 14
+wavs_audio1526985564579_2_6.wav 20
+wavs_audio1526985568317_6_4.wav 35
+wavs_audio1526985574403_2_7.wav 20
+wavs_audio1526985577889_1_2.wav 15
+wavs_audio1526985586019_4_3.wav 18
+wavs_audio1526985604764_3_3.wav 15
+wavs_audio1526985612198_4_4.wav 12

exp/asr_stats_raw_en_word/logdir/stats.11/valid/batch_keys

@@ -0,0 +1,2 @@
+speech
+text

exp/asr_stats_raw_en_word/logdir/stats.11/valid/speech_shape

@@ -0,0 +1,24 @@
+wavs_audio1526984388383_2_1.wav 65280
+wavs_audio1526984404711_2_5.wav 33600
+wavs_audio1526984413796_6_4.wav 96960
+wavs_audio1526984415584_5_3.wav 44160
+wavs_audio1526984449871_3_1.wav 43200
+wavs_audio1526984474571_4_1.wav 56640
+wavs_audio1526984496503_3_4.wav 35520
+wavs_audio1526984588328_5_4.wav 65280
+wavs_audio1526984612454_5_6.wav 55680
+wavs_audio1526984792401_6_4.wav 77760
+wavs_audio1526984800745_4_4.wav 40320
+wavs_audio1526984841890_2_2.wav 72960
+wavs_audio1526984842433_1_6.wav 49920
+wavs_audio1526984971124_3_1.wav 66240
+wavs_audio1526984975810_1_2.wav 55680
+wavs_audio1526984977437_2_4.wav 51840
+wavs_audio1526984992628_1_4.wav 41280
+wavs_audio1526985012725_3_4.wav 60480
+wavs_audio1526985045071_3_1.wav 46080
+wavs_audio1526985090396_2_5.wav 29760
+wavs_audio1526985126422_2_5.wav 47040
+wavs_audio1526985156008_4_5.wav 66240
+wavs_audio1526985267535_1_3.wav 57600
+wavs_audio1526985281042_4_4.wav 20160

exp/asr_stats_raw_en_word/logdir/stats.11/valid/stats_keys

@@ -0,0 +1,2 @@
+feats
+feats_lengths

exp/asr_stats_raw_en_word/logdir/stats.11/valid/text_shape

@@ -0,0 +1,24 @@
+wavs_audio1526984388383_2_1.wav 23
+wavs_audio1526984404711_2_5.wav 14
+wavs_audio1526984413796_6_4.wav 35
+wavs_audio1526984415584_5_3.wav 29
+wavs_audio1526984449871_3_1.wav 15
+wavs_audio1526984474571_4_1.wav 18
+wavs_audio1526984496503_3_4.wav 13
+wavs_audio1526984588328_5_4.wav 29
+wavs_audio1526984612454_5_6.wav 27
+wavs_audio1526984792401_6_4.wav 35
+wavs_audio1526984800745_4_4.wav 12
+wavs_audio1526984841890_2_2.wav 23
+wavs_audio1526984842433_1_6.wav 11
+wavs_audio1526984971124_3_1.wav 15
+wavs_audio1526984975810_1_2.wav 15
+wavs_audio1526984977437_2_4.wav 21
+wavs_audio1526984992628_1_4.wav 20
+wavs_audio1526985012725_3_4.wav 13
+wavs_audio1526985045071_3_1.wav 15
+wavs_audio1526985090396_2_5.wav 14
+wavs_audio1526985126422_2_5.wav 14
+wavs_audio1526985156008_4_5.wav 20
+wavs_audio1526985267535_1_3.wav 9
+wavs_audio1526985281042_4_4.wav 12

exp/asr_stats_raw_en_word/logdir/stats.12.log

@@ -0,0 +1,495 @@
+# Running on r099.ib.bridges2.psc.edu
+# Started at Tue Dec 21 22:25:07 EST 2021
+# SLURMD_NODENAME=r099
+# SLURM_ARRAY_JOB_ID=5730432
+# SLURM_ARRAY_TASK_COUNT=32
+# SLURM_ARRAY_TASK_ID=12
+# SLURM_ARRAY_TASK_MAX=32
+# SLURM_ARRAY_TASK_MIN=1
+# SLURM_ARRAY_TASK_STEP=1
+# SLURM_CLUSTER_NAME=bridges2
+# SLURM_CONF=/var/spool/slurm/d/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=1
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GTIDS=0
+# SLURM_JOBID=5730465
+# SLURM_JOB_ACCOUNT=cis210027p
+# SLURM_JOB_CPUS_PER_NODE=1
+# SLURM_JOB_GID=24886
+# SLURM_JOB_ID=5730465
+# SLURM_JOB_NAME=stats.sh
+# SLURM_JOB_NODELIST=r099
+# SLURM_JOB_NUM_NODES=1
+# SLURM_JOB_PARTITION=RM-shared
+# SLURM_JOB_QOS=rm
+# SLURM_JOB_UID=82326
+# SLURM_JOB_USER=ganesank
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_CPU=2000
+# SLURM_NNODES=1
+# SLURM_NODEID=0
+# SLURM_NODELIST=r099
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/ocean/projects/cis210027p/ganesank/karthik_new/espnet/egs2/sinhala/asr1
+# SLURM_SUBMIT_HOST=br012.ib.bridges2.psc.edu
+# SLURM_TASKS_PER_NODE=1
+# SLURM_TASK_PID=7310
+# SLURM_TOPOLOGY_ADDR=r099
+# SLURM_TOPOLOGY_ADDR_PATTERN=node
+# SLURM_WORKING_CLUSTER=bridges2:br003:6814:9216:109
+# python3 -m espnet2.bin.asr_train --collect_stats true --use_preprocessor true --bpemodel none --token_type word --token_list data/en_token_list/word/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --train_data_path_and_name_and_type dump/raw/train/wav.scp,speech,sound --train_data_path_and_name_and_type dump/raw/train/text,text,text --valid_data_path_and_name_and_type dump/raw/valid/wav.scp,speech,sound --valid_data_path_and_name_and_type dump/raw/valid/text,text,text --train_shape_file exp/asr_stats_raw_en_word/logdir/train.12.scp --valid_shape_file exp/asr_stats_raw_en_word/logdir/valid.12.scp --output_dir exp/asr_stats_raw_en_word/logdir/stats.12 --config conf/train_asr.yaml --frontend_conf fs=16k 
+/ocean/projects/cis210027p/ganesank/karthik_new/espnet/tools/venv/bin/python3 /ocean/projects/cis210027p/ganesank/karthik_new/espnet/espnet2/bin/asr_train.py --collect_stats true --use_preprocessor true --bpemodel none --token_type word --token_list data/en_token_list/word/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --train_data_path_and_name_and_type dump/raw/train/wav.scp,speech,sound --train_data_path_and_name_and_type dump/raw/train/text,text,text --valid_data_path_and_name_and_type dump/raw/valid/wav.scp,speech,sound --valid_data_path_and_name_and_type dump/raw/valid/text,text,text --train_shape_file exp/asr_stats_raw_en_word/logdir/train.12.scp --valid_shape_file exp/asr_stats_raw_en_word/logdir/valid.12.scp --output_dir exp/asr_stats_raw_en_word/logdir/stats.12 --config conf/train_asr.yaml --frontend_conf fs=16k
+[r099] 2021-12-21 22:25:10,190 (asr:382) INFO: Vocabulary size: 40
+[r099] 2021-12-21 22:25:10,529 (abs_task:1132) INFO: pytorch.version=1.8.1+cu102, cuda.available=False, cudnn.version=7605, cudnn.benchmark=False, cudnn.deterministic=True
+[r099] 2021-12-21 22:25:10,534 (abs_task:1133) INFO: Model structure:
+ESPnetASRModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=512, hop_length=128, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (time_warp): TimeWarp(window=5, mode=bicubic)
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 30], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxis(mask_width_range=[0, 40], num_mask=2, axis=time)
+  )
+  (normalize): UtteranceMVN(norm_means=True, norm_vars=False)
+  (encoder): TransformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 256, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=4864, out_features=256, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (9): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (10): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (11): EncoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(40, 256)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=256, out_features=40, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=256, out_features=256, bias=True)
+          (linear_k): Linear(in_features=256, out_features=256, bias=True)
+          (linear_v): Linear(in_features=256, out_features=256, bias=True)
+          (linear_out): Linear(in_features=256, out_features=256, bias=True)
+          (dropout): Dropout(p=0.0, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=256, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=256, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((256,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=256, out_features=40, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetASRModel
+    Total Number of model parameters: 27.12 M
+    Number of trainable parameters: 27.12 M (100.0%)
+    Size: 108.49 MB
+    Type: torch.float32
+[r099] 2021-12-21 22:25:10,534 (abs_task:1136) INFO: Optimizer:
+Adam (
+Parameter Group 0
+    amsgrad: False
+    betas: (0.9, 0.999)
+    eps: 1e-08
+    initial_lr: 0.0002
+    lr: 8e-09
+    weight_decay: 0
+)
+[r099] 2021-12-21 22:25:10,534 (abs_task:1137) INFO: Scheduler: WarmupLR(warmup_steps=25000)
+[r099] 2021-12-21 22:25:10,536 (abs_task:1146) INFO: Saving the configuration in exp/asr_stats_raw_en_word/logdir/stats.12/config.yaml
+[r099] 2021-12-21 22:25:10,546 (abs_task:1157) INFO: Namespace(config='conf/train_asr.yaml', print_config=False, log_level='INFO', dry_run=False, iterator_type='sequence', output_dir='exp/asr_stats_raw_en_word/logdir/stats.12', ngpu=0, seed=0, num_workers=1, num_att_plot=3, dist_backend='nccl', dist_init_method='env://', dist_world_size=None, dist_rank=None, local_rank=None, dist_master_addr=None, dist_master_port=None, dist_launcher=None, multiprocessing_distributed=False, unused_parameters=False, sharded_ddp=False, cudnn_enabled=True, cudnn_benchmark=False, cudnn_deterministic=True, collect_stats=True, write_collected_feats=False, max_epoch=50, patience=None, val_scheduler_criterion=('valid', 'loss'), early_stopping_criterion=('valid', 'loss', 'min'), best_model_criterion=[('train', 'loss', 'min'), ('valid', 'loss', 'min'), ('train', 'acc', 'max'), ('valid', 'acc', 'max')], keep_nbest_models=5, grad_clip=5.0, grad_clip_type=2.0, grad_noise=False, accum_grad=1, no_forward_run=False, resume=False, train_dtype='float32', use_amp=False, log_interval=None, use_tensorboard=True, use_wandb=False, wandb_project=None, wandb_id=None, wandb_entity=None, wandb_name=None, wandb_model_log_interval=-1, detect_anomaly=False, pretrain_path=None, init_param=[], ignore_init_mismatch=False, freeze_param=[], num_iters_per_epoch=None, batch_size=20, valid_batch_size=None, batch_bins=1000000, valid_batch_bins=None, train_shape_file=['exp/asr_stats_raw_en_word/logdir/train.12.scp'], valid_shape_file=['exp/asr_stats_raw_en_word/logdir/valid.12.scp'], batch_type='folded', valid_batch_type=None, fold_length=[], sort_in_batch='descending', sort_batch='descending', multiple_iterator=False, chunk_length=500, chunk_shift_ratio=0.5, num_cache_chunks=1024, train_data_path_and_name_and_type=[('dump/raw/train/wav.scp', 'speech', 'sound'), ('dump/raw/train/text', 'text', 'text')], valid_data_path_and_name_and_type=[('dump/raw/valid/wav.scp', 'speech', 'sound'), ('dump/raw/valid/text', 'text', 'text')], allow_variable_data_keys=False, max_cache_size=0.0, max_cache_fd=32, valid_max_cache_size=None, optim='adam', optim_conf={'lr': 0.0002}, scheduler='warmuplr', scheduler_conf={'warmup_steps': 25000}, token_list=['<blank>', '<unk>', '්', 'න', 'ම', 'ක', 'ල', 'ි', 'ු', 'ග', 'ේ', 'ර', 'ත', 'ද', 'ව', 'ට', 'ඕ', 'ී', 'ප', 'ය', 'ෙ', 'ස', 'ණ', 'ා', 'ැ', 'RequestAcc.balance', 'Moneywithdraw', 'Moneydeposit', 'Moneytransfer', 'Billpayments', 'බ', 'ඉ', 'ශ', 'ෂ', 'ඩ', 'Creditcardpayments', 'එ', '\u200d', 'හ', '<sos/eos>'], init=None, input_size=None, ctc_conf={'dropout_rate': 0.0, 'ctc_type': 'builtin', 'reduce': True, 'ignore_nan_grad': True}, model_conf={'ctc_weight': 0.5, 'ignore_id': -1, 'lsm_weight': 0.0, 'length_normalized_loss': False, 'report_cer': True, 'report_wer': True, 'sym_space': '<space>', 'sym_blank': '<blank>', 'extract_feats_in_collect_stats': True}, use_preprocessor=True, token_type='word', bpemodel=None, non_linguistic_symbols=None, cleaner=None, g2p=None, speech_volume_normalize=None, rir_scp=None, rir_apply_prob=1.0, noise_scp=None, noise_apply_prob=1.0, noise_db_range='13_15', frontend='default', frontend_conf={'fs': '16k'}, specaug='specaug', specaug_conf={'apply_time_warp': True, 'time_warp_window': 5, 'time_warp_mode': 'bicubic', 'apply_freq_mask': True, 'freq_mask_width_range': [0, 30], 'num_freq_mask': 2, 'apply_time_mask': True, 'time_mask_width_range': [0, 40], 'num_time_mask': 2}, normalize='utterance_mvn', normalize_conf={}, preencoder=None, preencoder_conf={}, encoder='transformer', encoder_conf={'output_size': 256, 'attention_heads': 4, 'linear_units': 2048, 'num_blocks': 12, 'dropout_rate': 0.1, 'positional_dropout_rate': 0.1, 'attention_dropout_rate': 0.0, 'input_layer': 'conv2d', 'normalize_before': True}, postencoder=None, postencoder_conf={}, decoder='transformer', decoder_conf={'attention_heads': 4, 'linear_units': 2048, 'num_blocks': 6, 'dropout_rate': 0.1, 'positional_dropout_rate': 0.1, 'self_attention_dropout_rate': 0.0, 'src_attention_dropout_rate': 0.0}, required=['output_dir', 'token_list'], version='0.10.3a3', distributed=False)
+# Accounting: begin_time=1640143507
+# Accounting: end_time=1640143519
+# Accounting: time=12 threads=1
+# Finished at Tue Dec 21 22:25:19 EST 2021 with status 0

exp/asr_stats_raw_en_word/logdir/stats.12/config.yaml

@@ -0,0 +1,231 @@
+config: conf/train_asr.yaml
+print_config: false
+log_level: INFO
+dry_run: false
+iterator_type: sequence
+output_dir: exp/asr_stats_raw_en_word/logdir/stats.12
+ngpu: 0
+seed: 0
+num_workers: 1
+num_att_plot: 3
+dist_backend: nccl
+dist_init_method: env://
+dist_world_size: null
+dist_rank: null
+local_rank: null
+dist_master_addr: null
+dist_master_port: null
+dist_launcher: null
+multiprocessing_distributed: false
+unused_parameters: false
+sharded_ddp: false
+cudnn_enabled: true
+cudnn_benchmark: false
+cudnn_deterministic: true
+collect_stats: true
+write_collected_feats: false
+max_epoch: 50
+patience: null
+val_scheduler_criterion:
+- valid
+- loss
+early_stopping_criterion:
+- valid
+- loss
+- min
+best_model_criterion:
+-   - train
+    - loss
+    - min
+-   - valid
+    - loss
+    - min
+-   - train
+    - acc
+    - max
+-   - valid
+    - acc
+    - max
+keep_nbest_models: 5
+grad_clip: 5.0
+grad_clip_type: 2.0
+grad_noise: false
+accum_grad: 1
+no_forward_run: false
+resume: false
+train_dtype: float32
+use_amp: false
+log_interval: null
+use_tensorboard: true
+use_wandb: false
+wandb_project: null
+wandb_id: null
+wandb_entity: null
+wandb_name: null
+wandb_model_log_interval: -1
+detect_anomaly: false
+pretrain_path: null
+init_param: []
+ignore_init_mismatch: false
+freeze_param: []
+num_iters_per_epoch: null
+batch_size: 20
+valid_batch_size: null
+batch_bins: 1000000
+valid_batch_bins: null
+train_shape_file:
+- exp/asr_stats_raw_en_word/logdir/train.12.scp
+valid_shape_file:
+- exp/asr_stats_raw_en_word/logdir/valid.12.scp
+batch_type: folded
+valid_batch_type: null
+fold_length: []
+sort_in_batch: descending
+sort_batch: descending
+multiple_iterator: false
+chunk_length: 500
+chunk_shift_ratio: 0.5
+num_cache_chunks: 1024
+train_data_path_and_name_and_type:
+-   - dump/raw/train/wav.scp
+    - speech
+    - sound
+-   - dump/raw/train/text
+    - text
+    - text
+valid_data_path_and_name_and_type:
+-   - dump/raw/valid/wav.scp
+    - speech
+    - sound
+-   - dump/raw/valid/text
+    - text
+    - text
+allow_variable_data_keys: false
+max_cache_size: 0.0
+max_cache_fd: 32
+valid_max_cache_size: null
+optim: adam
+optim_conf:
+    lr: 0.0002
+scheduler: warmuplr
+scheduler_conf:
+    warmup_steps: 25000
+token_list:
+- <blank>
+- <unk>
+- ්
+- න
+- ම
+- ක
+- ල
+- ි
+- ු
+- ග
+- ේ
+- ර
+- ත
+- ද
+- ව
+- ට
+- ඕ
+- ී
+- ප
+- ය
+- ෙ
+- ස
+- ණ
+- ා
+- ැ
+- RequestAcc.balance
+- Moneywithdraw
+- Moneydeposit
+- Moneytransfer
+- Billpayments
+- බ
+- ඉ
+- ශ
+- ෂ
+- ඩ
+- Creditcardpayments
+- එ
+- ‍
+- හ
+- <sos/eos>
+init: null
+input_size: null
+ctc_conf:
+    dropout_rate: 0.0
+    ctc_type: builtin
+    reduce: true
+    ignore_nan_grad: true
+model_conf:
+    ctc_weight: 0.5
+    ignore_id: -1
+    lsm_weight: 0.0
+    length_normalized_loss: false
+    report_cer: true
+    report_wer: true
+    sym_space: <space>
+    sym_blank: <blank>
+    extract_feats_in_collect_stats: true
+use_preprocessor: true
+token_type: word
+bpemodel: null
+non_linguistic_symbols: null
+cleaner: null
+g2p: null
+speech_volume_normalize: null
+rir_scp: null
+rir_apply_prob: 1.0
+noise_scp: null
+noise_apply_prob: 1.0
+noise_db_range: '13_15'
+frontend: default
+frontend_conf:
+    fs: 16k
+specaug: specaug
+specaug_conf:
+    apply_time_warp: true
+    time_warp_window: 5
+    time_warp_mode: bicubic
+    apply_freq_mask: true
+    freq_mask_width_range:
+    - 0
+    - 30
+    num_freq_mask: 2
+    apply_time_mask: true
+    time_mask_width_range:
+    - 0
+    - 40
+    num_time_mask: 2
+normalize: utterance_mvn
+normalize_conf: {}
+preencoder: null
+preencoder_conf: {}
+encoder: transformer
+encoder_conf:
+    output_size: 256
+    attention_heads: 4
+    linear_units: 2048
+    num_blocks: 12
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    attention_dropout_rate: 0.0
+    input_layer: conv2d
+    normalize_before: true
+postencoder: null
+postencoder_conf: {}
+decoder: transformer
+decoder_conf:
+    attention_heads: 4
+    linear_units: 2048
+    num_blocks: 6
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    self_attention_dropout_rate: 0.0
+    src_attention_dropout_rate: 0.0
+required:
+- output_dir
+- token_list
+version: 0.10.3a3
+distributed: false

exp/asr_stats_raw_en_word/logdir/stats.12/train/batch_keys

@@ -0,0 +1,2 @@
+speech
+text

exp/asr_stats_raw_en_word/logdir/stats.12/train/speech_shape

@@ -0,0 +1,191 @@
+wavs_audio1526985624260_3_5.wav 39360
+wavs_audio1526985631794_3_6.wav 48960
+wavs_audio1526985634840_1_5.wav 65280
+wavs_audio1526985637609_3_7.wav 25920
+wavs_audio1526985654320_3_8.wav 29760
+wavs_audio1526985665007_2_1.wav 56640
+wavs_audio1526985684367_2_2.wav 59520
+wavs_audio1526985684560_4_1.wav 29760
+wavs_audio1526985699919_2_4.wav 69120
+wavs_audio1526985720271_2_6.wav 50880
+wavs_audio1526985748846_3_1.wav 50880
+wavs_audio1526985771667_3_2.wav 32640
+wavs_audio1526985788422_3_3.wav 54720
+wavs_audio1526985811423_3_5.wav 39360
+wavs_audio1526985842900_4_1.wav 41280
+wavs_audio1526985843874_1_5.wav 35520
+wavs_audio1526985854679_1_6.wav 29760
+wavs_audio1526985856978_4_2.wav 41280
+wavs_audio1526985863962_4_3.wav 40320
+wavs_audio1526985866725_1_7.wav 53760
+wavs_audio1526985867124_1_1.wav 60480
+wavs_audio1526985870791_4_4.wav 31680
+wavs_audio1526985880824_4_5.wav 49920
+wavs_audio1526985884692_2_1.wav 35520
+wavs_audio1526985891709_2_2.wav 34560
+wavs_audio1526985919844_5_2.wav 98880
+wavs_audio1526985929522_2_3.wav 43200
+wavs_audio1526985937315_2_4.wav 36480
+wavs_audio1526985944403_2_5.wav 29760
+wavs_audio1526985946985_1_1.wav 46080
+wavs_audio1526985964774_1_2.wav 46080
+wavs_audio1526985970398_1_4.wav 84480
+wavs_audio1526985973382_3_1.wav 26880
+wavs_audio1526985977416_1_1.wav 48960
+wavs_audio1526985989468_3_3.wav 28800
+wavs_audio1526985993949_5_4.wav 48000
+wavs_audio1526985996680_3_4.wav 28800
+wavs_audio1526985999822_1_3.wav 54720
+wavs_audio1526986001645_1_2.wav 52800
+wavs_audio1526986004390_5_5.wav 65280
+wavs_audio1526986006027_3_5.wav 30720
+wavs_audio1526986013569_3_6.wav 32640
+wavs_audio1526986013869_5_6.wav 60480
+wavs_audio1526986017131_1_1.wav 69120
+wavs_audio1526986023134_3_7.wav 28800
+wavs_audio1526986026460_1_4.wav 57600
+wavs_audio1526986028850_1_4.wav 61440
+wavs_audio1526986031364_3_8.wav 30720
+wavs_audio1526986038758_1_5.wav 56640
+wavs_audio1526986045359_1_3.wav 42240
+wavs_audio1526986049128_1_6.wav 48000
+wavs_audio1526986054679_1_4.wav 53760
+wavs_audio1526986056001_1_6.wav 54720
+wavs_audio1526986058107_4_2.wav 36480
+wavs_audio1526986060927_6_2.wav 50880
+wavs_audio1526986062560_1_7.wav 70080
+wavs_audio1526986065613_4_3.wav 27840
+wavs_audio1526986070442_4_4.wav 24000
+wavs_audio1526986073227_1_5.wav 48960
+wavs_audio1526986074610_1_2.wav 51840
+wavs_audio1526986075386_1_7.wav 82560
+wavs_audio1526986094591_5_1.wav 43200
+wavs_audio1526986098023_1_8.wav 81600
+wavs_audio1526986100533_1_3.wav 39360
+wavs_audio1526986101739_2_4.wav 61440
+wavs_audio1526986106691_1_6.wav 45120
+wavs_audio1526986110995_1_1.wav 32640
+wavs_audio1526986112701_2_5.wav 47040
+wavs_audio1526986114724_5_3.wav 36480
+wavs_audio1526986134094_5_5.wav 37440
+wavs_audio1526986134721_1_7.wav 65280
+wavs_audio1526986140693_2_7.wav 48960
+wavs_audio1526986142358_5_6.wav 42240
+wavs_audio1526986145427_1_3.wav 31680
+wavs_audio1526986147839_1_8.wav 65280
+wavs_audio1526986150434_1_6.wav 37440
+wavs_audio1526986156067_2_1.wav 42240
+wavs_audio1526986160452_3_1.wav 52800
+wavs_audio1526986161162_1_4.wav 48960
+wavs_audio1526986168606_2_1.wav 48000
+wavs_audio1526986172550_1_7.wav 87360
+wavs_audio1526986181104_2_2.wav 54720
+wavs_audio1526986186429_1_6.wav 36480
+wavs_audio1526986189325_2_3.wav 44160
+wavs_audio1526986193798_2_2.wav 55680
+wavs_audio1526986199038_2_4.wav 49920
+wavs_audio1526986200639_1_7.wav 60480
+wavs_audio1526986212315_2_5.wav 47040
+wavs_audio1526986213279_1_8.wav 60480
+wavs_audio1526986216840_5_7.wav 32640
+wavs_audio1526986220541_2_6.wav 46080
+wavs_audio1526986237311_2_1.wav 50880
+wavs_audio1526986243940_2_4.wav 80640
+wavs_audio1526986252397_3_2.wav 43200
+wavs_audio1526986253549_1_1.wav 58560
+wavs_audio1526986258147_2_5.wav 48960
+wavs_audio1526986258855_6_2.wav 36480
+wavs_audio1526986267624_2_2.wav 42240
+wavs_audio1526986273630_2_1.wav 49920
+wavs_audio1526986278268_2_3.wav 42240
+wavs_audio1526986282138_2_7.wav 60480
+wavs_audio1526986286341_6_3.wav 40320
+wavs_audio1526986291717_1_1.wav 42240
+wavs_audio1526986294920_3_6.wav 46080
+wavs_audio1526986295948_2_4.wav 38400
+wavs_audio1526986296671_2_3.wav 38400
+wavs_audio1526986300507_6_4.wav 47040
+wavs_audio1526986302631_3_7.wav 44160
+wavs_audio1526986308389_2_5.wav 38400
+wavs_audio1526986310297_3_1.wav 39360
+wavs_audio1526986311003_3_8.wav 51840
+wavs_audio1526986315765_1_1.wav 28800
+wavs_audio1526986321856_3_2.wav 48000
+wavs_audio1526986323874_4_1.wav 46080
+wavs_audio1526986328478_2_4.wav 40320
+wavs_audio1526986332943_4_2.wav 51840
+wavs_audio1526986334059_3_3.wav 72960
+wavs_audio1526986345036_2_7.wav 38400
+wavs_audio1526986346831_4_3.wav 53760
+wavs_audio1526986349586_2_5.wav 38400
+wavs_audio1526986356132_4_4.wav 43200
+wavs_audio1526986356658_3_5.wav 65280
+wavs_audio1526986363308_2_6.wav 36480
+wavs_audio1526986363441_3_2.wav 39360
+wavs_audio1526986364384_4_5.wav 53760
+wavs_audio1526986378346_2_7.wav 35520
+wavs_audio1526986380912_5_1.wav 64320
+wavs_audio1526986383681_1_3.wav 24960
+wavs_audio1526986384092_3_1.wav 37440
+wavs_audio1526986387611_3_4.wav 62400
+wavs_audio1526986389985_1_4.wav 33600
+wavs_audio1526986394615_3_7.wav 55680
+wavs_audio1526986395198_5_2.wav 64320
+wavs_audio1526986399200_3_2.wav 35520
+wavs_audio1526986399414_1_5.wav 28800
+wavs_audio1526986404099_3_8.wav 52800
+wavs_audio1526986404411_5_3.wav 58560
+wavs_audio1526986416370_3_4.wav 29760
+wavs_audio1526986428897_3_5.wav 33600
+wavs_audio1526986430267_3_6.wav 57600
+wavs_audio1526986431551_5_4.wav 57600
+wavs_audio1526986434577_4_1.wav 57600
+wavs_audio1526986435901_3_2.wav 48000
+wavs_audio1526986440680_5_5.wav 60480
+wavs_audio1526986443586_3_7.wav 38400
+wavs_audio1526986445561_3_7.wav 24960
+wavs_audio1526986449787_3_3.wav 80640
+wavs_audio1526986452840_3_8.wav 27840
+wavs_audio1526986453435_3_8.wav 42240
+wavs_audio1526986464469_5_7.wav 91200
+wavs_audio1526986464670_3_4.wav 40320
+wavs_audio1526986474896_1_1.wav 69120
+wavs_audio1526986477588_4_2.wav 37440
+wavs_audio1526986480308_3_5.wav 60480
+wavs_audio1526986481319_6_1.wav 60480
+wavs_audio1526986485336_4_3.wav 30720
+wavs_audio1526986492918_4_4.wav 26880
+wavs_audio1526986494268_3_6.wav 44160
+wavs_audio1526986498705_4_5.wav 33600
+wavs_audio1526986505536_4_3.wav 57600
+wavs_audio1526986514851_6_2.wav 53760
+wavs_audio1526986518301_5_1.wav 43200
+wavs_audio1526986520630_3_8.wav 34560
+wavs_audio1526986524204_6_3.wav 56640
+wavs_audio1526986528126_4_1.wav 42240
+wavs_audio1526986529822_4_5.wav 47040
+wavs_audio1526986534112_1_3.wav 39360
+wavs_audio1526986536567_6_4.wav 61440
+wavs_audio1526986539592_4_3.wav 41280
+wavs_audio1526986541667_4_2.wav 46080
+wavs_audio1526986556588_4_1.wav 31680
+wavs_audio1526986559296_1_4.wav 46080
+wavs_audio1526986567858_4_5.wav 47040
+wavs_audio1526986579068_4_4.wav 31680
+wavs_audio1526986584329_5_2.wav 48960
+wavs_audio1526986588425_1_6.wav 41280
+wavs_audio1526986590238_4_5.wav 52800
+wavs_audio1526986593494_5_1.wav 57600
+wavs_audio1526986599320_5_3.wav 49920
+wavs_audio1526986600296_5_3.wav 71040
+wavs_audio1526986607539_5_2.wav 56640
+wavs_audio1526986608700_5_4.wav 52800
+wavs_audio1526986613380_5_4.wav 64320
+wavs_audio1526986617030_5_1.wav 55680
+wavs_audio1526986623759_5_5.wav 54720
+wavs_audio1526986630976_5_5.wav 52800
+wavs_audio1526986631150_5_2.wav 60480
+wavs_audio1526986637317_5_6.wav 60480
+wavs_audio1526986644558_5_3.wav 55680
+wavs_audio1526986647579_5_7.wav 45120
+wavs_audio1526986649587_5_3.wav 98880