#!/bin/bash
  
  # Copyright 2012-2014  Johns Hopkins University (Author: Daniel Povey).
  #                2013  Xiaohui Zhang
  #                2013  Guoguo Chen
  #                2014  Vimal Manohar
  # Apache 2.0.
  
  
  # train_pnorm_simple2.sh is as train_pnorm_simple.sh but it uses the "new" egs
  # format, created by get_egs2.sh.
  
  # train_pnorm_simple.sh is a modified version of train_pnorm_fast.sh.  Like
  # train_pnorm_fast.sh, it uses the `online' preconditioning, which is faster
  # (especially on GPUs).  The difference is that the learning-rate schedule is
  # simpler, with the learning rate exponentially decreasing during training,
  # and no phase where the learning rate is constant.
  #
  # Also, the final model-combination is done a bit differently: we combine models
  # over typically a whole epoch, and because that would be too many iterations to
  # easily be able to combine over, we arrange the iterations into groups (20
  # groups by default) and average over each group.
  #
  # [Vimal Manohar - Oct 2014]
  # The script now supports realignment during training, which can be done by
  # specifying realign_epochs.
  
  # Begin configuration section.
  cmd=run.pl
  num_epochs=15      # Number of epochs of training;
                     # the number of iterations is worked out from this.
  initial_learning_rate=0.04
  final_learning_rate=0.004
  bias_stddev=0.5
  pnorm_input_dim=3000
  pnorm_output_dim=300
  p=2
  presoftmax_prior_scale_power=-0.25 # use the specified power value on the priors (inverse priors)
                                     # to scale the pre-softmax outputs
  minibatch_size=128 # by default use a smallish minibatch size for neural net
                     # training; this controls instability which would otherwise
                     # be a problem with multi-threaded update.
  
  samples_per_iter=400000 # each iteration of training, see this many samples
                          # per job.  This option is passed to get_egs.sh
  num_jobs_nnet=4    # Number of neural net jobs to run in parallel.  This option
                     # is passed to get_egs.sh.
  prior_subset_size=10000 # 10k samples per job, for computing priors.  Should be
                          # more than enough.
  num_jobs_compute_prior=10 # these are single-threaded, run on CPU.
  get_egs_stage=0
  online_ivector_dir=
  
  
  max_models_combine=20 # The "max_models_combine" is the maximum number of models we give
    # to the final 'combine' stage, but these models will themselves be averages of
    # iteration-number ranges.
  
  shuffle_buffer_size=5000 # This "buffer_size" variable controls randomization of the samples
                  # on each iter.  You could set it to 0 or to a large value for complete
                  # randomization, but this would both consume memory and cause spikes in
                  # disk I/O.  Smaller is easier on disk and memory but less random.  It's
                  # not a huge deal though, as samples are anyway randomized right at the start.
                  # (the point of this is to get data in different minibatches on different iterations,
                  # since in the preconditioning method, 2 samples in the same minibatch can
                  # affect each others' gradients.
  
  add_layers_period=2 # by default, add new layers every 2 iterations.
  num_hidden_layers=3
  stage=-4
  
  splice_width=4 # meaning +- 4 frames on each side for second LDA
  left_context= # if set, overrides splice-width
  right_context= # if set, overrides splice-width.
  randprune=4.0 # speeds up LDA.
  alpha=4.0 # relates to preconditioning.
  update_period=4 # relates to online preconditioning: says how often we update the subspace.
  num_samples_history=2000 # relates to online preconditioning
  max_change_per_sample=0.075
  precondition_rank_in=20  # relates to online preconditioning
  precondition_rank_out=80 # relates to online preconditioning
  
  mix_up=0 # Number of components to mix up to (should be > #tree leaves, if
          # specified.)
  num_threads=16
  parallel_opts="--num-threads 16 --mem 1G"
    # by default we use 16 threads; this lets the queue know.
    # note: parallel_opts doesn't automatically get adjusted if you adjust num-threads.
  combine_num_threads=8
  combine_parallel_opts="--num-threads 8"  # queue options for the "combine" stage.
  cleanup=true
  egs_dir=
  lda_opts=
  lda_dim=
  egs_opts=
  io_opts="--max-jobs-run 5" # for jobs with a lot of I/O, limits the number running at one time.
  transform_dir=     # If supplied, overrides alidir
  cmvn_opts=  # will be passed to get_lda.sh and get_egs.sh, if supplied.
              # only relevant for "raw" features, not lda.
  feat_type=  # Can be used to force "raw" features.
  align_cmd=              # The cmd that is passed to steps/nnet2/align.sh
  align_use_gpu=          # Passed to use_gpu in steps/nnet2/align.sh [yes/no]
  realign_epochs=         # List of epochs, the beginning of which realignment is done
  num_jobs_align=30       # Number of jobs for realignment
  # End configuration section.
  
  
  echo "$0 $@"  # Print the command line for logging
  
  if [ -f path.sh ]; then . ./path.sh; fi
  . parse_options.sh || exit 1;
  
  if [ $# != 4 ]; then
    echo "Usage: $0 [opts] <data> <lang> <ali-dir> <exp-dir>"
    echo " e.g.: $0 data/train data/lang exp/tri3_ali exp/tri4_nnet"
    echo ""
    echo "Main options (for others, see top of script file)"
    echo "  --config <config-file>                           # config file containing options"
    echo "  --cmd (utils/run.pl|utils/queue.pl <queue opts>) # how to run jobs."
    echo "  --num-epochs <#epochs|15>                        # Number of epochs of training"
    echo "  --initial-learning-rate <initial-learning-rate|0.02> # Learning rate at start of training, e.g. 0.02 for small"
    echo "                                                       # data, 0.01 for large data"
    echo "  --final-learning-rate  <final-learning-rate|0.004>   # Learning rate at end of training, e.g. 0.004 for small"
    echo "                                                   # data, 0.001 for large data"
    echo "  --num-hidden-layers <#hidden-layers|2>           # Number of hidden layers, e.g. 2 for 3 hours of data, 4 for 100hrs"
    echo "  --add-layers-period <#iters|2>                   # Number of iterations between adding hidden layers"
    echo "  --mix-up <#pseudo-gaussians|0>                   # This option now does nothing; please remove it."
    echo "  --presoftmax-prior-scale-power <power|-0.25>     # use the specified power value on the priors (inverse priors) "
    echo "                                                   # to scale the pre-softmax outputs."
    echo "                                                   # (set to 0.0 to disable the presoftmax element scale)"
    echo "  --num-jobs-nnet <num-jobs|8>                     # Number of parallel jobs to use for main neural net"
    echo "                                                   # training (will affect results as well as speed; try 8, 16)"
    echo "                                                   # Note: if you increase this, you may want to also increase"
    echo "                                                   # the learning rate."
    echo "  --num-threads <num-threads|16>                   # Number of parallel threads per job (will affect results"
    echo "                                                   # as well as speed; may interact with batch size; if you increase"
    echo "                                                   # this, you may want to decrease the batch size."
    echo "  --parallel-opts <opts|\"--num-threads 16 --mem 1G\">      # extra options to pass to e.g. queue.pl for processes that"
    echo "                                                   # use multiple threads... "
    echo "  --io-opts <opts|\"--max-jobs-run 10\">                      # Options given to e.g. queue.pl for jobs that do a lot of I/O."
    echo "  --minibatch-size <minibatch-size|128>            # Size of minibatch to process (note: product with --num-threads"
    echo "                                                   # should not get too large, e.g. >2k)."
    echo "  --samples-per-iter <#samples|400000>             # Number of samples of data to process per iteration, per"
    echo "                                                   # process."
    echo "  --splice-width <width|4>                         # Number of frames on each side to append for feature input"
    echo "                                                   # (note: we splice processed, typically 40-dimensional frames"
    echo "  --lda-dim <dim|250>                              # Dimension to reduce spliced features to with LDA"
    echo "  --realign-epochs <list-of-epochs|\"\">           # A list of space-separated epoch indices the beginning of which"
    echo "                                                   # realignment is to be done"
    echo "  --align-cmd (utils/run.pl|utils/queue.pl <queue opts>) # passed to align.sh"
    echo "  --align-use-gpu (yes/no)                         # specify is gpu is to be used for realignment"
    echo "  --num-jobs-align <#njobs|30>                     # Number of jobs to perform realignment"
    echo "  --stage <stage|-4>                               # Used to run a partially-completed training process from somewhere in"
    echo "                                                   # the middle."
  
  
    exit 1;
  fi
  
  data=$1
  lang=$2
  alidir=$3
  dir=$4
  
  if [ ! -z "$realign_epochs" ]; then
    [ -z "$align_cmd" ] && echo "$0: realign_epochs specified but align_cmd not specified" && exit 1
    [ -z "$align_use_gpu" ] && echo "$0: realign_epochs specified but align_use_gpu not specified" && exit 1
  fi
  
  # Check some files.
  for f in $data/feats.scp $lang/L.fst $alidir/ali.1.gz $alidir/final.mdl $alidir/tree; do
    [ ! -f $f ] && echo "$0: no such file $f" && exit 1;
  done
  
  
  # Set some variables.
  num_leaves=`tree-info $alidir/tree 2>/dev/null | grep num-pdfs | awk '{print $2}'` || exit 1
  [ -z $num_leaves ] && echo "\$num_leaves is unset" && exit 1
  [ "$num_leaves" -eq "0" ] && echo "\$num_leaves is 0" && exit 1
  
  nj=`cat $alidir/num_jobs` || exit 1;  # number of jobs in alignment dir...
  # in this dir we'll have just one job.
  sdata=$data/split$nj
  utils/split_data.sh $data $nj
  
  mkdir -p $dir/log
  echo $nj > $dir/num_jobs
  cp $alidir/tree $dir
  
  utils/lang/check_phones_compatible.sh $lang/phones.txt $alidir/phones.txt || exit 1;
  cp $lang/phones.txt $dir || exit 1;
  
  extra_opts=()
  [ ! -z "$cmvn_opts" ] && extra_opts+=(--cmvn-opts "$cmvn_opts")
  [ ! -z "$feat_type" ] && extra_opts+=(--feat-type $feat_type)
  [ ! -z "$online_ivector_dir" ] && extra_opts+=(--online-ivector-dir $online_ivector_dir)
  [ -z "$transform_dir" ] && transform_dir=$alidir
  extra_opts+=(--transform-dir $transform_dir)
  [ -z "$left_context" ] && left_context=$splice_width
  [ -z "$right_context" ] && right_context=$splice_width
  extra_opts+=(--left-context $left_context --right-context $right_context)
  
  if [ $stage -le -4 ]; then
    echo "$0: calling get_lda.sh"
    steps/nnet2/get_lda.sh $lda_opts "${extra_opts[@]}" --cmd "$cmd" $data $lang $alidir $dir || exit 1;
  fi
  
  # these files will have been written by get_lda.sh
  feat_dim=$(cat $dir/feat_dim) || exit 1;
  ivector_dim=$(cat $dir/ivector_dim) || exit 1;
  lda_dim=$(cat $dir/lda_dim) || exit 1;
  
  if [ $stage -le -3 ] && [ -z "$egs_dir" ]; then
    echo "$0: calling get_egs2.sh"
    steps/nnet2/get_egs2.sh $egs_opts "${extra_opts[@]}"  --io-opts "$io_opts" \
      --samples-per-iter $samples_per_iter --stage $get_egs_stage \
      --cmd "$cmd" $egs_opts $data $alidir $dir/egs || exit 1;
  fi
  
  if [ -z $egs_dir ]; then
    egs_dir=$dir/egs
  fi
  
  frames_per_eg=$(cat $egs_dir/info/frames_per_eg) || { echo "error: no such file $egs_dir/info/frames_per_eg"; exit 1; }
  num_archives=$(cat $egs_dir/info/num_archives) || { echo "error: no such file $egs_dir/info/frames_per_eg"; exit 1; }
  
  # num_archives_expanded considers each separate label-position from
  # 0..frames_per_eg-1 to be a separate archive.
  num_archives_expanded=$[$num_archives*$frames_per_eg]
  
  if [ $num_jobs_nnet -gt $num_archives_expanded ]; then
    echo "$0: --num-jobs-nnet cannot exceed num-archives*frames-per-eg which is $num_archives_expanded"
    echo "$0: setting --num-jobs-nnet to $num_archives_expanded"
    num_jobs_nnet=$num_archives_expanded
  fi
  
  if ! [ $num_hidden_layers -ge 1 ]; then
    echo "Invalid num-hidden-layers $num_hidden_layers"
    exit 1
  fi
  
  if [ $stage -le -2 ]; then
    echo "$0: initializing neural net";
    lda_mat=$dir/lda.mat
    tot_input_dim=$[$feat_dim+$ivector_dim]
  
    online_preconditioning_opts="alpha=$alpha num-samples-history=$num_samples_history update-period=$update_period rank-in=$precondition_rank_in rank-out=$precondition_rank_out max-change-per-sample=$max_change_per_sample"
  
    stddev=`perl -e "print 1.0/sqrt($pnorm_input_dim);"`
    cat >$dir/nnet.config <<EOF
  SpliceComponent input-dim=$tot_input_dim left-context=$left_context right-context=$right_context const-component-dim=$ivector_dim
  FixedAffineComponent matrix=$lda_mat
  AffineComponentPreconditionedOnline input-dim=$lda_dim output-dim=$pnorm_input_dim $online_preconditioning_opts learning-rate=$initial_learning_rate param-stddev=$stddev bias-stddev=$bias_stddev
  PnormComponent input-dim=$pnorm_input_dim output-dim=$pnorm_output_dim p=$p
  NormalizeComponent dim=$pnorm_output_dim
  AffineComponentPreconditionedOnline input-dim=$pnorm_output_dim output-dim=$num_leaves $online_preconditioning_opts learning-rate=$initial_learning_rate param-stddev=0 bias-stddev=0
  SoftmaxComponent dim=$num_leaves
  EOF
  
    # to hidden.config it will write the part of the config corresponding to a
    # single hidden layer; we need this to add new layers.
    cat >$dir/hidden.config <<EOF
  AffineComponentPreconditionedOnline input-dim=$pnorm_output_dim output-dim=$pnorm_input_dim $online_preconditioning_opts learning-rate=$initial_learning_rate param-stddev=$stddev bias-stddev=$bias_stddev
  PnormComponent input-dim=$pnorm_input_dim output-dim=$pnorm_output_dim p=$p
  NormalizeComponent dim=$pnorm_output_dim
  EOF
    $cmd $dir/log/nnet_init.log \
      nnet-am-init $alidir/tree $lang/topo "nnet-init $dir/nnet.config -|" \
      $dir/0.mdl || exit 1;
  fi
  
  if [ $stage -le -1 ]; then
    echo "Training transition probabilities and setting priors"
    $cmd $dir/log/train_trans.log \
      nnet-train-transitions $dir/0.mdl "ark:gunzip -c $alidir/ali.*.gz|" $dir/0.mdl \
      || exit 1;
  
    if [ "$presoftmax_prior_scale_power" != "0.0" ]; then
      echo "prepare vector assignment for FixedScaleComponent before softmax"
      echo "(use priors^$presoftmax_prior_scale_power and rescale to average 1)"
  
      # obtains raw pdf count
      $cmd JOB=1:$nj $dir/log/acc_pdf.JOB.log \
        ali-to-post "ark:gunzip -c $alidir/ali.JOB.gz|" ark:- \| \
        post-to-tacc --per-pdf=true --binary=false $alidir/final.mdl ark:- $dir/JOB.pacc || exit 1;
      cat $dir/*.pacc > $dir/pacc
      rm $dir/*.pacc
      awk -v power=$presoftmax_prior_scale_power \
        '{ for(i=2; i<=NF-1; i++) {sum[i]+=$i} }
        END {
          for (i=2; i<=NF-1; i++) {total+=sum[i]}
          ave_pdf=int(total/(NF-2)); total+=0.01*ave_pdf*(NF-2)
          for (i=2; i<=NF-1; i++) {rescale+=((sum[i]+0.01*ave_pdf)/total)^power}
          rescale/=(NF-2)
          printf " [ "; for (i=2; i<=NF-1; i++) {printf("%f ", ((sum[i]+0.01*ave_pdf)/total)^power/rescale)}; print "]"
        }' $dir/pacc > $dir/presoftmax_prior_scale_vecfile
  
      echo "FixedScaleComponent scales=$dir/presoftmax_prior_scale_vecfile" > $dir/per_element.config
      echo "insert an additional layer of FixedScaleComponent before softmax"
      inp=`nnet-am-info $dir/0.mdl | grep 'Softmax' | awk '{print $2}'`
      nnet-init $dir/per_element.config - | nnet-insert --insert-at=$inp --randomize-next-component=false $dir/0.mdl - $dir/0.mdl
    fi
  fi
  
  # set num_iters so that as close as possible, we process the data $num_epochs
  # times, i.e. $num_iters*$num_jobs_nnet == $num_epochs*$num_archives_expanded
  num_iters=$[($num_epochs*$num_archives_expanded)/$num_jobs_nnet]
  
  echo "$0: Will train for $num_epochs epochs = $num_iters iterations"
  echo "$0: Will not do mix up"
  
  finish_add_layers_iter=$[$num_hidden_layers * $add_layers_period]
  # This is when we decide to mix up from: halfway between when we've finished
  # adding the hidden layers and the end of training.
  mix_up_iter=$[($num_iters + $finish_add_layers_iter)/2]
  
  if [ $num_threads -eq 1 ]; then
    parallel_suffix="-simple" # this enables us to use GPU code if
                           # we have just one thread.
    parallel_train_opts=
    if ! cuda-compiled; then
      echo "$0: WARNING: you are running with one thread but you have not compiled"
      echo "   for CUDA.  You may be running a setup optimized for GPUs.  If you have"
      echo "   GPUs and have nvcc installed, go to src/ and do ./configure; make"
    fi
  else
    parallel_suffix="-parallel"
    parallel_train_opts="--num-threads=$num_threads"
  fi
  
  
  approx_iters_per_epoch=$[$num_iters/$num_epochs]
  # First work out how many models we want to combine over in the final
  # nnet-combine-fast invocation.  This equals
  # min(max(max_models_combine, iters_per_epoch),
  #     2/3 * iters_after_mixup)
  num_models_combine=$max_models_combine
  if [ $num_models_combine -lt $approx_iters_per_epoch ]; then
    num_models_combine=$approx_iters_per_epoch
  fi
  iters_after_mixup_23=$[(($num_iters-$mix_up_iter-1)*2)/3]
  if [ $num_models_combine -gt $iters_after_mixup_23 ]; then
    num_models_combine=$iters_after_mixup_23
  fi
  first_model_combine=$[$num_iters-$num_models_combine+1]
  
  x=0
  
  for realign_epoch in $realign_epochs; do
    # compare the equation below with the equation we use to set num_iters above.
    # note, realign_epochs may be floating-point, which is why we don't use $[] to
    # do the math.
    realign_iter=$(perl -e 'print int(($ARGV[0]*$ARGV[1])/$ARGV[2]);' $realign_epoch $num_archives_expanded $num_jobs_nnet)
    realign_this_iter[$realign_iter]=$realign_epoch
  done
  
  cur_egs_dir=$egs_dir
  
  while [ $x -lt $num_iters ]; do
  
    if [ ! -z "${realign_this_iter[$x]}" ]; then
      prev_egs_dir=$cur_egs_dir
      cur_egs_dir=$dir/egs_${realign_this_iter[$x]}
    fi
  
    if [ $x -ge 0 ] && [ $stage -le $x ]; then
      if [ ! -z "${realign_this_iter[$x]}" ]; then
        epoch=${realign_this_iter[$x]}
  
  
  
        echo "Getting average posterior for purposes of adjusting the priors."
        # Note: this just uses CPUs, using a smallish subset of data.
        # always use the first egs archive, which makes the script simpler;
        # we're using different random subsets of it.
        rm $dir/post.$x.*.vec 2>/dev/null
        $cmd JOB=1:$num_jobs_compute_prior $dir/log/get_post.$x.JOB.log \
          nnet-copy-egs --srand=JOB --frame=random ark:$prev_egs_dir/egs.1.ark ark:- \| \
          nnet-subset-egs --srand=JOB --n=$prior_subset_size ark:- ark:- \| \
          nnet-compute-from-egs "nnet-to-raw-nnet $dir/$x.mdl -|" ark:- ark:- \| \
          matrix-sum-rows ark:- ark:- \| vector-sum ark:- $dir/post.$x.JOB.vec || exit 1;
  
        sleep 3;  # make sure there is time for $dir/post.$x.*.vec to appear.
  
        $cmd $dir/log/vector_sum.$x.log \
          vector-sum $dir/post.$x.*.vec $dir/post.$x.vec || exit 1;
        rm $dir/post.$x.*.vec;
  
        echo "Re-adjusting priors based on computed posteriors"
        $cmd $dir/log/adjust_priors.$x.log \
          nnet-adjust-priors $dir/$x.mdl $dir/post.$x.vec $dir/$x.mdl || exit 1;
  
        sleep 2
  
        steps/nnet2/align.sh --nj $num_jobs_align --cmd "$align_cmd" --use-gpu $align_use_gpu \
          --transform-dir "$transform_dir" --online-ivector-dir "$online_ivector_dir" \
          --iter $x $data $lang $dir $dir/ali_$epoch || exit 1
  
        steps/nnet2/relabel_egs2.sh --cmd "$cmd" --iter $x $dir/ali_$epoch \
          $prev_egs_dir $cur_egs_dir || exit 1
  
        if $cleanup && [[ $prev_egs_dir =~ $dir/egs* ]]; then
          steps/nnet2/remove_egs.sh $prev_egs_dir
        fi
      fi
  
      # Set off jobs doing some diagnostics, in the background.
      # Use the egs dir from the previous iteration for the diagnostics
      $cmd $dir/log/compute_prob_valid.$x.log \
        nnet-compute-prob $dir/$x.mdl ark:$cur_egs_dir/valid_diagnostic.egs &
      $cmd $dir/log/compute_prob_train.$x.log \
        nnet-compute-prob $dir/$x.mdl ark:$cur_egs_dir/train_diagnostic.egs &
      if [ $x -gt 0 ] && [ ! -f $dir/log/mix_up.$[$x-1].log ]; then
        $cmd $dir/log/progress.$x.log \
          nnet-show-progress --use-gpu=no $dir/$[$x-1].mdl $dir/$x.mdl \
          ark:$cur_egs_dir/train_diagnostic.egs '&&' \
          nnet-am-info $dir/$x.mdl &
      fi
  
      echo "Training neural net (pass $x)"
  
      if [ $x -gt 0 ] && \
        [ $x -le $[($num_hidden_layers-1)*$add_layers_period] ] && \
        [ $[($x-1) % $add_layers_period] -eq 0 ]; then
  
        inp=`nnet-am-info $dir/$x.mdl | grep 'Softmax' | awk '{print $2}'`
        if [ "$presoftmax_prior_scale_power" != "0.0" ]; then
          inp=$[$inp-2]
        else
          inp=$[$inp-1]
        fi
        mdl="nnet-init --srand=$x $dir/hidden.config - | nnet-insert --insert-at=$inp $dir/$x.mdl - - |"
      else
        mdl=$dir/$x.mdl
      fi
      if [ $x -eq 0 ] || [ "$mdl" != "$dir/$x.mdl" ]; then
        # on iteration zero or when we just added a layer, use a smaller minibatch
        # size and just one job: the model-averaging doesn't seem to be helpful
        # when the model is changing too fast (i.e. it worsens the objective
        # function), and the smaller minibatch size will help to keep
        # the update stable.
        this_minibatch_size=$[$minibatch_size/2];
        do_average=false
      else
        this_minibatch_size=$minibatch_size
        do_average=true
      fi
  
      rm $dir/.error 2>/dev/null
  
  
      ( # this sub-shell is so that when we "wait" below,
        # we only wait for the training jobs that we just spawned,
        # not the diagnostic jobs that we spawned above.
  
        # We can't easily use a single parallel SGE job to do the main training,
        # because the computation of which archive and which --frame option
        # to use for each job is a little complex, so we spawn each one separately.
        for n in $(seq $num_jobs_nnet); do
          k=$[$x*$num_jobs_nnet + $n - 1]; # k is a zero-based index that we'll derive
                                           # the other indexes from.
          archive=$[($k%$num_archives)+1]; # work out the 1-based archive index.
          frame=$[(($k/$num_archives)%$frames_per_eg)]; # work out the 0-based frame
          # index; this increases more slowly than the archive index because the
          # same archive with different frame indexes will give similar gradients,
          # so we want to separate them in time.
  
          $cmd $parallel_opts $dir/log/train.$x.$n.log \
            nnet-train$parallel_suffix $parallel_train_opts \
            --minibatch-size=$this_minibatch_size --srand=$x "$mdl" \
            "ark,bg:nnet-copy-egs --frame=$frame ark:$cur_egs_dir/egs.$archive.ark ark:-|nnet-shuffle-egs --buffer-size=$shuffle_buffer_size --srand=$x ark:- ark:-|" \
            $dir/$[$x+1].$n.mdl || touch $dir/.error &
        done
        wait
      )
      # the error message below is not that informative, but $cmd will
      # have printed a more specific one.
      [ -f $dir/.error ] && echo "$0: error on iteration $x of training" && exit 1;
  
      nnets_list=
      for n in `seq 1 $num_jobs_nnet`; do
        nnets_list="$nnets_list $dir/$[$x+1].$n.mdl"
      done
  
      learning_rate=`perl -e '($x,$n,$i,$f)=@ARGV; print ($x >= $n ? $f : $i*exp($x*log($f/$i)/$n));' $[$x+1] $num_iters $initial_learning_rate $final_learning_rate`;
  
      if $do_average; then
        # average the output of the different jobs.
        $cmd $dir/log/average.$x.log \
          nnet-am-average $nnets_list - \| \
          nnet-am-copy --learning-rate=$learning_rate - $dir/$[$x+1].mdl || exit 1;
      else
        # choose the best from the different jobs.
        n=$(perl -e '($nj,$pat)=@ARGV; $best_n=1; $best_logprob=-1.0e+10; for ($n=1;$n<=$nj;$n++) {
            $fn = sprintf($pat,$n); open(F, "<$fn") || die "Error opening log file $fn";
            undef $logprob; while (<F>) { if (m/log-prob-per-frame=(\S+)/) { $logprob=$1; } }
            close(F); if (defined $logprob && $logprob > $best_logprob) { $best_logprob=$logprob;
            $best_n=$n; } } print "$best_n
  "; ' $num_jobs_nnet $dir/log/train.$x.%d.log) || exit 1;
        [ -z "$n" ] && echo "Error getting best model" && exit 1;
        $cmd $dir/log/select.$x.log \
          nnet-am-copy --learning-rate=$learning_rate $dir/$[$x+1].$n.mdl $dir/$[$x+1].mdl || exit 1;
      fi
  
      if [ "$mix_up" -gt 0 ] && [ $x -eq $mix_up_iter ]; then
        echo "Warning: the mix up opertion is disabled!"
        echo "    Ignore mix up leaves number specified"
      fi
      rm $nnets_list
      [ ! -f $dir/$[$x+1].mdl ] && exit 1;
      if [ -f $dir/$[$x-1].mdl ] && $cleanup && \
         [ $[($x-1)%100] -ne 0  ] && [ $[$x-1] -lt $first_model_combine ]; then
        rm $dir/$[$x-1].mdl
      fi
    fi
    x=$[$x+1]
  done
  
  
  if [ $stage -le $num_iters ]; then
    echo "Doing final combination to produce final.mdl"
  
    # Now do combination.
    nnets_list=()
    # the if..else..fi statement below sets 'nnets_list'.
    if [ $max_models_combine -lt $num_models_combine ]; then
      # The number of models to combine is too large, e.g. > 20.  In this case,
      # each argument to nnet-combine-fast will be an average of multiple models.
      cur_offset=0 # current offset from first_model_combine.
      for n in $(seq $max_models_combine); do
        next_offset=$[($n*$num_models_combine)/$max_models_combine]
        sub_list=""
        for o in $(seq $cur_offset $[$next_offset-1]); do
          iter=$[$first_model_combine+$o]
          mdl=$dir/$iter.mdl
          [ ! -f $mdl ] && echo "Expected $mdl to exist" && exit 1;
          sub_list="$sub_list $mdl"
        done
        nnets_list[$[$n-1]]="nnet-am-average $sub_list - |"
        cur_offset=$next_offset
      done
    else
      nnets_list=
      for n in $(seq 0 $[num_models_combine-1]); do
        iter=$[$first_model_combine+$n]
        mdl=$dir/$iter.mdl
        [ ! -f $mdl ] && echo "Expected $mdl to exist" && exit 1;
        nnets_list[$n]=$mdl
      done
    fi
  
  
    # Below, use --use-gpu=no to disable nnet-combine-fast from using a GPU, as
    # if there are many models it can give out-of-memory error; set num-threads to 8
    # to speed it up (this isn't ideal...)
    num_egs=`nnet-copy-egs ark:$cur_egs_dir/combine.egs ark:/dev/null 2>&1 | tail -n 1 | awk '{print $NF}'`
    mb=$[($num_egs+$combine_num_threads-1)/$combine_num_threads]
    [ $mb -gt 512 ] && mb=512
    # Setting --initial-model to a large value makes it initialize the combination
    # with the average of all the models.  It's important not to start with a
    # single model, or, due to the invariance to scaling that these nonlinearities
    # give us, we get zero diagonal entries in the fisher matrix that
    # nnet-combine-fast uses for scaling, which after flooring and inversion, has
    # the effect that the initial model chosen gets much higher learning rates
    # than the others.  This prevents the optimization from working well.
    $cmd $combine_parallel_opts $dir/log/combine.log \
      nnet-combine-fast --initial-model=100000 --num-lbfgs-iters=40 --use-gpu=no \
        --num-threads=$combine_num_threads \
        --verbose=3 --minibatch-size=$mb "${nnets_list[@]}" ark:$cur_egs_dir/combine.egs \
        $dir/final.mdl || exit 1;
  
    # Normalize stddev for affine or block affine layers that are followed by a
    # pnorm layer and then a normalize layer.
    $cmd $dir/log/normalize.log \
      nnet-normalize-stddev $dir/final.mdl $dir/final.mdl || exit 1;
  
    # Compute the probability of the final, combined model with
    # the same subset we used for the previous compute_probs, as the
    # different subsets will lead to different probs.
    $cmd $dir/log/compute_prob_valid.final.log \
      nnet-compute-prob $dir/final.mdl ark:$cur_egs_dir/valid_diagnostic.egs &
    $cmd $dir/log/compute_prob_train.final.log \
      nnet-compute-prob $dir/final.mdl ark:$cur_egs_dir/train_diagnostic.egs &
  fi
  
  if [ $stage -le $[$num_iters+1] ]; then
    echo "Getting average posterior for purposes of adjusting the priors."
    # Note: this just uses CPUs, using a smallish subset of data.
    rm $dir/post.$x.*.vec 2>/dev/null
    $cmd JOB=1:$num_jobs_compute_prior $dir/log/get_post.$x.JOB.log \
      nnet-copy-egs --frame=random --srand=JOB ark:$cur_egs_dir/egs.1.ark ark:- \| \
      nnet-subset-egs --srand=JOB --n=$prior_subset_size ark:- ark:- \| \
      nnet-compute-from-egs "nnet-to-raw-nnet $dir/final.mdl -|" ark:- ark:- \| \
      matrix-sum-rows ark:- ark:- \| vector-sum ark:- $dir/post.$x.JOB.vec || exit 1;
  
    sleep 3;  # make sure there is time for $dir/post.$x.*.vec to appear.
  
    $cmd $dir/log/vector_sum.$x.log \
     vector-sum $dir/post.$x.*.vec $dir/post.$x.vec || exit 1;
  
    rm $dir/post.$x.*.vec;
  
    echo "Re-adjusting priors based on computed posteriors"
    $cmd $dir/log/adjust_priors.final.log \
      nnet-adjust-priors $dir/final.mdl $dir/post.$x.vec $dir/final.mdl || exit 1;
  fi
  
  
  if [ ! -f $dir/final.mdl ]; then
    echo "$0: $dir/final.mdl does not exist."
    # we don't want to clean up if the training didn't succeed.
    exit 1;
  fi
  
  sleep 2
  
  echo Done
  
  if $cleanup; then
    echo Cleaning up data
    if [[ $cur_egs_dir =~ $dir/egs* ]]; then
      steps/nnet2/remove_egs.sh $cur_egs_dir
    fi
  
    echo Removing most of the models
    for x in `seq 0 $num_iters`; do
      if [ $[$x%100] -ne 0 ] && [ $x -ne $num_iters ] && [ -f $dir/$x.mdl ]; then
         # delete all but every 100th model; don't delete the ones which combine to form the final model.
        rm $dir/$x.mdl
      fi
    done
  fi