#!/bin/bash
  
  # This factorized TDNN (TDNN-F) script is adapted from SWBD recipe 7q.
  # It uses resnet-style skip connections.
  # For details, refer to the paper:
  # "Semi-Orthogonal Low-Rank Matrix Factorization for Deep Neural Networks", Daniel Povey, Gaofeng Cheng, Yiming Wang, Ke Li, Hainan Xu, Mahsa Yarmohamadi, Sanjeev Khudanpur, Interspeech 2018
  
  # %WER 70.27 [ 41375 / 58881, 3487 ins, 22831 del, 15057 sub ] exp/chain_train_worn_simu_u400k_cleaned_rvb/tdnn1b_sp/decode_dev_beamformit_dereverb_ref_2stage/wer_12_0.0
  # %WER 70.28 [ 41383 / 58881, 4486 ins, 19616 del, 17281 sub ] exp/chain_train_worn_simu_u400k_cleaned_rvb/tdnn1b_sp/decode_dev_beamformit_ref_2stage/wer_11_0.0
  # %WER 72.62 [ 42761 / 58881, 4545 ins, 21618 del, 16598 sub ] exp/chain_train_worn_simu_u400k_cleaned_rvb/tdnn1b_sp/decode_dev_beamformit_ref/wer_11_0.0
  # %WER 72.64 [ 42772 / 58881, 4556 ins, 21618 del, 16598 sub ] exp/chain_train_worn_simu_u400k_cleaned_rvb/tdnn1b_sp/decode_dev_beamformit_dereverb_ref/wer_11_0.0
  
  # steps/info/chain_dir_info.pl exp/chain_train_worn_simu_u400k_cleaned_rvb/tdnn_1b_sp
  # exp/chain_train_worn_simu_u400k_cleaned_rvb/tdnn1b_sp/: num-iters=317 nj=3..16 num-params=17.0M dim=40+100->2792 combine=-0.149->-0.149 (over 2) xent:train/valid[210,316,final]=(-2.50,-1.99,-2.00/-2.36,-1.95,-1.95) logprob:train/valid[210,316,final]=(-0.228,-0.136,-0.136/-0.223,-0.156,-0.155)
  
  set -e
  
  # configs for 'chain'
  stage=0
  nj=96
  train_set=train_worn_u400k
  test_sets="dev_worn dev_beamformit_ref"
  gmm=tri3
  nnet3_affix=_train_worn_u400k
  lm_suffix=
  
  # The rest are configs specific to this script.  Most of the parameters
  # are just hardcoded at this level, in the commands below.
  affix=1b   # affix for the TDNN directory name
  tree_affix=
  train_stage=-10
  get_egs_stage=-10
  decode_iter=
  
  num_epochs=4
  common_egs_dir=
  # training options
  # training chunk-options
  chunk_width=140,100,160
  xent_regularize=0.1
  dropout_schedule='0,0@0.20,0.5@0.50,0'
  
  # training options
  srand=0
  remove_egs=true
  
  #decode options
  test_online_decoding=false  # if true, it will run the last decoding stage.
  
  # End configuration section.
  echo "$0 $@"  # Print the command line for logging
  
  . ./cmd.sh
  . ./path.sh
  . ./utils/parse_options.sh
  
  if ! cuda-compiled; then
    cat <<EOF && exit 1
  This script is intended to be used with GPUs but you have not compiled Kaldi with CUDA
  If you want to use GPUs (and have them), go to src/, and configure and make on a machine
  where "nvcc" is installed.
  EOF
  fi
  
  # The iVector-extraction and feature-dumping parts are the same as the standard
  # nnet3 setup, and you can skip them by setting "--stage 11" if you have already
  # run those things.
  local/nnet3/run_ivector_common.sh --stage $stage \
                                    --train-set $train_set \
                                    --test-sets "$test_sets" \
                                    --gmm $gmm \
                                    --nnet3-affix "$nnet3_affix" || exit 1;
  
  # Problem: We have removed the "train_" prefix of our training set in
  # the alignment directory names! Bad!
  gmm_dir=exp/$gmm
  tree_dir=exp/chain${nnet3_affix}/tree_sp${tree_affix:+_$tree_affix}
  lang=data/lang_chain
  lat_dir=exp/chain${nnet3_affix}/${gmm}_${train_set}_sp_lats
  dir=exp/chain${nnet3_affix}/tdnn${affix}_sp
  train_data_dir=data/${train_set}_sp_hires
  lores_train_data_dir=data/${train_set}_sp
  train_ivector_dir=exp/nnet3${nnet3_affix}/ivectors_${train_set}_sp_hires
  
  for f in $gmm_dir/final.mdl $train_data_dir/feats.scp $train_ivector_dir/ivector_online.scp \
      $lores_train_data_dir/feats.scp; do
    [ ! -f $f ] && echo "$0: expected file $f to exist" && exit 1
  done
  
  if [ $stage -le 10 ]; then
    echo "$0: creating lang directory $lang with chain-type topology"
    # Create a version of the lang/ directory that has one state per phone in the
    # topo file. [note, it really has two states.. the first one is only repeated
    # once, the second one has zero or more repeats.]
    if [ -d $lang ]; then
      if [ $lang/L.fst -nt data/lang/L.fst ]; then
        echo "$0: $lang already exists, not overwriting it; continuing"
      else
        echo "$0: $lang already exists and seems to be older than data/lang..."
        echo " ... not sure what to do.  Exiting."
        exit 1;
      fi
    else
      cp -r data/lang $lang
      silphonelist=$(cat $lang/phones/silence.csl) || exit 1;
      nonsilphonelist=$(cat $lang/phones/nonsilence.csl) || exit 1;
      # Use our special topology... note that later on may have to tune this
      # topology.
      steps/nnet3/chain/gen_topo.py $nonsilphonelist $silphonelist >$lang/topo
    fi
  fi
  
  if [ $stage -le 11 ]; then
    # Get the alignments as lattices (gives the chain training more freedom).
    # use the same num-jobs as the alignments
    steps/align_fmllr_lats.sh --nj ${nj} --cmd "$train_cmd" --generate-ali-from-lats true \
      ${lores_train_data_dir} \
      data/lang $gmm_dir $lat_dir
    rm $lat_dir/fsts.*.gz # save space
  fi
  
  if [ $stage -le 12 ]; then
    # Build a tree using our new topology.  We know we have alignments for the
    # speed-perturbed data (local/nnet3/run_ivector_common.sh made them), so use
    # those.  The num-leaves is always somewhat less than the num-leaves from
    # the GMM baseline.
    if [ -f $tree_dir/final.mdl ]; then
       echo "$0: $tree_dir/final.mdl already exists, refusing to overwrite it."
       exit 1;
    fi
    steps/nnet3/chain/build_tree.sh \
      --frame-subsampling-factor 3 \
      --cmd "$train_cmd" 3500 ${lores_train_data_dir} \
      $lang $lat_dir $tree_dir
  fi
  
  if [ $stage -le 13 ]; then
    mkdir -p $dir
    echo "$0: creating neural net configs using the xconfig parser";
  
    num_targets=$(tree-info $tree_dir/tree |grep num-pdfs|awk '{print $2}')
    learning_rate_factor=$(echo "print 0.5/$xent_regularize" | python)
    affine_opts="l2-regularize=0.01 dropout-proportion=0.0 dropout-per-dim=true dropout-per-dim-continuous=true"
    tdnnf_opts="l2-regularize=0.01 dropout-proportion=0.0 bypass-scale=0.66"
    linear_opts="l2-regularize=0.01 orthonormal-constraint=-1.0"
    prefinal_opts="l2-regularize=0.01"
    output_opts="l2-regularize=0.002"
  
    mkdir -p $dir/configs
    cat <<EOF > $dir/configs/network.xconfig
    input dim=100 name=ivector
    input dim=40 name=input
  
    # please note that it is important to have input layer with the name=input
    # as the layer immediately preceding the fixed-affine-layer to enable
    # the use of short notation for the descriptor
    fixed-affine-layer name=lda input=Append(-1,0,1,ReplaceIndex(ivector, t, 0)) affine-transform-file=$dir/configs/lda.mat
  
    # the first splicing is moved before the lda layer, so no splicing here
    relu-batchnorm-dropout-layer name=tdnn1 $affine_opts dim=1536
    tdnnf-layer name=tdnnf2 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=1
    tdnnf-layer name=tdnnf3 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=1
    tdnnf-layer name=tdnnf4 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=1
    tdnnf-layer name=tdnnf5 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=0
    tdnnf-layer name=tdnnf6 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
    tdnnf-layer name=tdnnf7 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
    tdnnf-layer name=tdnnf8 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
    tdnnf-layer name=tdnnf9 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
    tdnnf-layer name=tdnnf10 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
    tdnnf-layer name=tdnnf11 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
    tdnnf-layer name=tdnnf12 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
    tdnnf-layer name=tdnnf13 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
    tdnnf-layer name=tdnnf14 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
    tdnnf-layer name=tdnnf15 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
    linear-component name=prefinal-l dim=256 $linear_opts
  
    prefinal-layer name=prefinal-chain input=prefinal-l $prefinal_opts big-dim=1536 small-dim=256
    output-layer name=output include-log-softmax=false dim=$num_targets $output_opts
  
    prefinal-layer name=prefinal-xent input=prefinal-l $prefinal_opts big-dim=1536 small-dim=256
    output-layer name=output-xent dim=$num_targets learning-rate-factor=$learning_rate_factor $output_opts
  EOF
    steps/nnet3/xconfig_to_configs.py --xconfig-file $dir/configs/network.xconfig --config-dir $dir/configs/
  fi
  
  if [ $stage -le 14 ]; then
    if [[ $(hostname -f) == *.clsp.jhu.edu ]] && [ ! -d $dir/egs/storage ]; then
      utils/create_split_dir.pl \
       /export/b0{3,4,5,6}/$USER/kaldi-data/egs/chime5-$(date +'%m_%d_%H_%M')/s5/$dir/egs/storage $dir/egs/storage
    fi
  
    steps/nnet3/chain/train.py --stage $train_stage \
      --cmd "$train_cmd --mem 4G" \
      --feat.online-ivector-dir=$train_ivector_dir \
      --feat.cmvn-opts "--norm-means=false --norm-vars=false" \
      --chain.xent-regularize $xent_regularize \
      --chain.leaky-hmm-coefficient 0.1 \
      --chain.l2-regularize 0.0 \
      --chain.apply-deriv-weights false \
      --chain.lm-opts="--num-extra-lm-states=2000" \
      --trainer.dropout-schedule "$dropout_schedule" \
      --trainer.add-option="--optimization.memory-compression-level=2" \
      --egs.dir "$common_egs_dir" \
      --egs.stage $get_egs_stage \
      --egs.opts "--frames-overlap-per-eg 0" \
      --egs.chunk-width $chunk_width \
      --trainer.num-chunk-per-minibatch 64 \
      --trainer.frames-per-iter 1500000 \
      --trainer.num-epochs $num_epochs \
      --trainer.optimization.num-jobs-initial 3 \
      --trainer.optimization.num-jobs-final 16 \
      --trainer.optimization.initial-effective-lrate 0.00025 \
      --trainer.optimization.final-effective-lrate 0.000025 \
      --trainer.max-param-change 2.0 \
      --cleanup.remove-egs $remove_egs \
      --feat-dir=$train_data_dir \
      --tree-dir=$tree_dir \
      --lat-dir=$lat_dir \
      --dir $dir  || exit 1;
  
  fi
  
  if [ $stage -le 15 ]; then
    # Note: it's not important to give mkgraph.sh the lang directory with the
    # matched topology (since it gets the topology file from the model).
    utils/mkgraph.sh \
      --self-loop-scale 1.0 data/lang${lm_suffix}/ \
      $tree_dir $tree_dir/graph${lm_suffix} || exit 1;
  fi
  
  if [ $stage -le 16 ]; then
    frames_per_chunk=$(echo $chunk_width | cut -d, -f1)
    rm $dir/.error 2>/dev/null || true
  
    for data in $test_sets; do
      (
        steps/nnet3/decode.sh \
            --acwt 1.0 --post-decode-acwt 10.0 \
            --frames-per-chunk $frames_per_chunk \
            --nj 8 --cmd "$decode_cmd"  --num-threads 4 \
            --online-ivector-dir exp/nnet3${nnet3_affix}/ivectors_${data}_hires \
            $tree_dir/graph${lm_suffix} data/${data}_hires ${dir}/decode${lm_suffix}_${data} || exit 1
      ) || touch $dir/.error &
    done
    wait
    [ -f $dir/.error ] && echo "$0: there was a problem while decoding" && exit 1
  fi
  
  exit 0;