============================================================================

                    MMI + SGMM2 Training & Decoding

============================================================================

steps/align_sgmm2.sh --nj 30 --cmd run.pl --mem 4G --transform-dir exp/tri3_ali --use-graphs true --use-gselect true data/train data/lang exp/sgmm2_4 exp/sgmm2_4_ali

steps/align_sgmm2.sh: feature type is lda

steps/align_sgmm2.sh: using transforms from exp/tri3_ali

steps/align_sgmm2.sh: aligning data in data/train using model exp/sgmm2_4/final.alimdl

steps/align_sgmm2.sh: computing speaker vectors (1st pass)

steps/align_sgmm2.sh: computing speaker vectors (2nd pass)

steps/align_sgmm2.sh: doing final alignment.

steps/align_sgmm2.sh: done aligning data.

steps/diagnostic/analyze_alignments.sh --cmd run.pl --mem 4G data/lang exp/sgmm2_4_ali

steps/diagnostic/analyze_alignments.sh: see stats in exp/sgmm2_4_ali/log/analyze_alignments.log

steps/make_denlats_sgmm2.sh --nj 30 --sub-split 30 --acwt 0.2 --lattice-beam 10.0 --beam 18.0 --cmd run.pl --mem 4G --transform-dir exp/tri3_ali data/train data/lang exp/sgmm2_4_ali exp/sgmm2_4_denlats

steps/make_denlats_sgmm2.sh: Making unigram grammar FST in exp/sgmm2_4_denlats/lang

steps/make_denlats_sgmm2.sh: Compiling decoding graph in exp/sgmm2_4_denlats/dengraph

tree-info exp/sgmm2_4_ali/tree

tree-info exp/sgmm2_4_ali/tree

fsttablecompose exp/sgmm2_4_denlats/lang/L_disambig.fst exp/sgmm2_4_denlats/lang/G.fst

fstdeterminizestar --use-log=true

fstminimizeencoded

fstpushspecial

fstisstochastic exp/sgmm2_4_denlats/lang/tmp/LG.fst

1.27271e-05 1.27271e-05

fstcomposecontext --context-size=3 --central-position=1 --read-disambig-syms=exp/sgmm2_4_denlats/lang/phones/disambig.int --write-disambig-syms=exp/sgmm2_4_denlats/lang/tmp/disambig_ilabels_3_1.int exp/sgmm2_4_denlats/lang/tmp/ilabels_3_1.7854

fstisstochastic exp/sgmm2_4_denlats/lang/tmp/CLG_3_1.fst

1.27657e-05 0

make-h-transducer --disambig-syms-out=exp/sgmm2_4_denlats/dengraph/disambig_tid.int --transition-scale=1.0 exp/sgmm2_4_denlats/lang/tmp/ilabels_3_1 exp/sgmm2_4_ali/tree exp/sgmm2_4_ali/final.mdl

fstdeterminizestar --use-log=true

fstrmsymbols exp/sgmm2_4_denlats/dengraph/disambig_tid.int

fsttablecompose exp/sgmm2_4_denlats/dengraph/Ha.fst exp/sgmm2_4_denlats/lang/tmp/CLG_3_1.fst

fstrmepslocal

fstminimizeencoded

fstisstochastic exp/sgmm2_4_denlats/dengraph/HCLGa.fst

0.000484977 -0.000485819

add-self-loops --self-loop-scale=0.1 --reorder=true exp/sgmm2_4_ali/final.mdl

steps/make_denlats_sgmm2.sh: feature type is lda

steps/make_denlats_sgmm2.sh: using fMLLR transforms from exp/tri3_ali

steps/make_denlats_sgmm2.sh: Merging archives for data subset 1

steps/make_denlats_sgmm2.sh: Merging archives for data subset 2

steps/make_denlats_sgmm2.sh: Merging archives for data subset 3

steps/make_denlats_sgmm2.sh: Merging archives for data subset 4

steps/make_denlats_sgmm2.sh: Merging archives for data subset 5

steps/make_denlats_sgmm2.sh: Merging archives for data subset 6

steps/make_denlats_sgmm2.sh: Merging archives for data subset 7

steps/make_denlats_sgmm2.sh: Merging archives for data subset 8

steps/make_denlats_sgmm2.sh: Merging archives for data subset 9

steps/make_denlats_sgmm2.sh: Merging archives for data subset 10

steps/make_denlats_sgmm2.sh: Merging archives for data subset 11

steps/make_denlats_sgmm2.sh: Merging archives for data subset 12

steps/make_denlats_sgmm2.sh: Merging archives for data subset 13

steps/make_denlats_sgmm2.sh: Merging archives for data subset 14

steps/make_denlats_sgmm2.sh: Merging archives for data subset 15

steps/make_denlats_sgmm2.sh: Merging archives for data subset 16

steps/make_denlats_sgmm2.sh: Merging archives for data subset 17

steps/make_denlats_sgmm2.sh: Merging archives for data subset 18

steps/make_denlats_sgmm2.sh: Merging archives for data subset 19

steps/make_denlats_sgmm2.sh: Merging archives for data subset 20

steps/make_denlats_sgmm2.sh: Merging archives for data subset 21

steps/make_denlats_sgmm2.sh: Merging archives for data subset 22

steps/make_denlats_sgmm2.sh: Merging archives for data subset 23

steps/make_denlats_sgmm2.sh: Merging archives for data subset 24

steps/make_denlats_sgmm2.sh: Merging archives for data subset 25

steps/make_denlats_sgmm2.sh: Merging archives for data subset 26

steps/make_denlats_sgmm2.sh: Merging archives for data subset 27

steps/make_denlats_sgmm2.sh: Merging archives for data subset 28

steps/make_denlats_sgmm2.sh: Merging archives for data subset 29

steps/make_denlats_sgmm2.sh: Merging archives for data subset 30

steps/make_denlats_sgmm2.sh: done generating denominator lattices with SGMMs.

steps/train_mmi_sgmm2.sh --acwt 0.2 --cmd run.pl --mem 4G --transform-dir exp/tri3_ali --boost 0.1 --drop-frames true data/train data/lang exp/sgmm2_4_ali exp/sgmm2_4_denlats exp/sgmm2_4_mmi_b0.1

steps/train_mmi_sgmm2.sh: feature type is lda

steps/train_mmi_sgmm2.sh: using transforms from exp/tri3_ali

steps/train_mmi_sgmm2.sh: using speaker vectors from exp/sgmm2_4_ali

steps/train_mmi_sgmm2.sh: using Gaussian-selection info from exp/sgmm2_4_ali

Iteration 0 of MMI training

Iteration 0: objf was 0.501479915681143, MMI auxf change was 0.0162951148758516

Iteration 1 of MMI training

Iteration 1: objf was 0.516025726621521, MMI auxf change was 0.00230327793795113

Iteration 2 of MMI training

Iteration 2: objf was 0.518545043633159, MMI auxf change was 0.000600702510528323

Iteration 3 of MMI training

Iteration 3: objf was 0.51940583564136, MMI auxf change was 0.000356484531344043

MMI training finished

steps/decode_sgmm2_rescore.sh --cmd run.pl --mem 4G --iter 1 --transform-dir exp/tri3/decode_dev data/lang_test_bg data/dev exp/sgmm2_4/decode_dev exp/sgmm2_4_mmi_b0.1/decode_dev_it1

steps/decode_sgmm2_rescore.sh: using speaker vectors from exp/sgmm2_4/decode_dev

steps/decode_sgmm2_rescore.sh: feature type is lda

steps/decode_sgmm2_rescore.sh: using transforms from exp/tri3/decode_dev

steps/decode_sgmm2_rescore.sh: rescoring lattices with SGMM model in exp/sgmm2_4_mmi_b0.1/1.mdl

steps/decode_sgmm2_rescore.sh --cmd run.pl --mem 4G --iter 1 --transform-dir exp/tri3/decode_test data/lang_test_bg data/test exp/sgmm2_4/decode_test exp/sgmm2_4_mmi_b0.1/decode_test_it1

steps/decode_sgmm2_rescore.sh: using speaker vectors from exp/sgmm2_4/decode_test

steps/decode_sgmm2_rescore.sh: feature type is lda

steps/decode_sgmm2_rescore.sh: using transforms from exp/tri3/decode_test

steps/decode_sgmm2_rescore.sh: rescoring lattices with SGMM model in exp/sgmm2_4_mmi_b0.1/1.mdl

steps/decode_sgmm2_rescore.sh --cmd run.pl --mem 4G --iter 2 --transform-dir exp/tri3/decode_dev data/lang_test_bg data/dev exp/sgmm2_4/decode_dev exp/sgmm2_4_mmi_b0.1/decode_dev_it2

steps/decode_sgmm2_rescore.sh: using speaker vectors from exp/sgmm2_4/decode_dev

steps/decode_sgmm2_rescore.sh: feature type is lda

steps/decode_sgmm2_rescore.sh: using transforms from exp/tri3/decode_dev

steps/decode_sgmm2_rescore.sh: rescoring lattices with SGMM model in exp/sgmm2_4_mmi_b0.1/2.mdl

steps/decode_sgmm2_rescore.sh --cmd run.pl --mem 4G --iter 2 --transform-dir exp/tri3/decode_test data/lang_test_bg data/test exp/sgmm2_4/decode_test exp/sgmm2_4_mmi_b0.1/decode_test_it2

steps/decode_sgmm2_rescore.sh: using speaker vectors from exp/sgmm2_4/decode_test

steps/decode_sgmm2_rescore.sh: feature type is lda

steps/decode_sgmm2_rescore.sh: using transforms from exp/tri3/decode_test

steps/decode_sgmm2_rescore.sh: rescoring lattices with SGMM model in exp/sgmm2_4_mmi_b0.1/2.mdl

steps/decode_sgmm2_rescore.sh --cmd run.pl --mem 4G --iter 3 --transform-dir exp/tri3/decode_dev data/lang_test_bg data/dev exp/sgmm2_4/decode_dev exp/sgmm2_4_mmi_b0.1/decode_dev_it3

steps/decode_sgmm2_rescore.sh: using speaker vectors from exp/sgmm2_4/decode_dev

steps/decode_sgmm2_rescore.sh: feature type is lda

steps/decode_sgmm2_rescore.sh: using transforms from exp/tri3/decode_dev

steps/decode_sgmm2_rescore.sh: rescoring lattices with SGMM model in exp/sgmm2_4_mmi_b0.1/3.mdl

steps/decode_sgmm2_rescore.sh --cmd run.pl --mem 4G --iter 3 --transform-dir exp/tri3/decode_test data/lang_test_bg data/test exp/sgmm2_4/decode_test exp/sgmm2_4_mmi_b0.1/decode_test_it3

steps/decode_sgmm2_rescore.sh: using speaker vectors from exp/sgmm2_4/decode_test

steps/decode_sgmm2_rescore.sh: feature type is lda

steps/decode_sgmm2_rescore.sh: using transforms from exp/tri3/decode_test

steps/decode_sgmm2_rescore.sh: rescoring lattices with SGMM model in exp/sgmm2_4_mmi_b0.1/3.mdl

steps/decode_sgmm2_rescore.sh --cmd run.pl --mem 4G --iter 4 --transform-dir exp/tri3/decode_dev data/lang_test_bg data/dev exp/sgmm2_4/decode_dev exp/sgmm2_4_mmi_b0.1/decode_dev_it4

steps/decode_sgmm2_rescore.sh: using speaker vectors from exp/sgmm2_4/decode_dev

steps/decode_sgmm2_rescore.sh: feature type is lda

steps/decode_sgmm2_rescore.sh: using transforms from exp/tri3/decode_dev

steps/decode_sgmm2_rescore.sh: rescoring lattices with SGMM model in exp/sgmm2_4_mmi_b0.1/4.mdl

steps/decode_sgmm2_rescore.sh --cmd run.pl --mem 4G --iter 4 --transform-dir exp/tri3/decode_test data/lang_test_bg data/test exp/sgmm2_4/decode_test exp/sgmm2_4_mmi_b0.1/decode_test_it4

steps/decode_sgmm2_rescore.sh: using speaker vectors from exp/sgmm2_4/decode_test

steps/decode_sgmm2_rescore.sh: feature type is lda

steps/decode_sgmm2_rescore.sh: using transforms from exp/tri3/decode_test

steps/decode_sgmm2_rescore.sh: rescoring lattices with SGMM model in exp/sgmm2_4_mmi_b0.1/4.mdl

============================================================================

                    DNN Hybrid Training & Decoding

============================================================================

steps/nnet2/train_tanh.sh --mix-up 5000 --initial-learning-rate 0.015 --final-learning-rate 0.002 --num-hidden-layers 2 --num-jobs-nnet 30 --cmd run.pl --mem 4G data/train data/lang exp/tri3_ali exp/tri4_nnet

steps/nnet2/train_tanh.sh: calling get_lda.sh

steps/nnet2/get_lda.sh --transform-dir exp/tri3_ali --splice-width 4 --cmd run.pl --mem 4G data/train data/lang exp/tri3_ali exp/tri4_nnet

steps/nnet2/get_lda.sh: feature type is lda

steps/nnet2/get_lda.sh: using transforms from exp/tri3_ali

feat-to-dim 'ark,s,cs:utils/subset_scp.pl --quiet 333 data/train/split30/1/feats.scp | apply-cmvn  --utt2spk=ark:data/train/split30/1/utt2spk scp:data/train/split30/1/cmvn.scp scp:- ark:- | splice-feats --left-context=3 --right-context=3 ark:- ark:- | transform-feats exp/tri4_nnet/final.mat ark:- ark:- | transform-feats --utt2spk=ark:data/train/split30/1/utt2spk ark:exp/tri3_ali/trans.1 ark:- ark:- |' -

transform-feats --utt2spk=ark:data/train/split30/1/utt2spk ark:exp/tri3_ali/trans.1 ark:- ark:-

apply-cmvn --utt2spk=ark:data/train/split30/1/utt2spk scp:data/train/split30/1/cmvn.scp scp:- ark:-

splice-feats --left-context=3 --right-context=3 ark:- ark:-

transform-feats exp/tri4_nnet/final.mat ark:- ark:-

WARNING (feat-to-dim[5.2.124~1396-70748]:Close():kaldi-io.cc:501) Pipe utils/subset_scp.pl --quiet 333 data/train/split30/1/feats.scp | apply-cmvn  --utt2spk=ark:data/train/split30/1/utt2spk scp:data/train/split30/1/cmvn.scp scp:- ark:- | splice-feats --left-context=3 --right-context=3 ark:- ark:- | transform-feats exp/tri4_nnet/final.mat ark:- ark:- | transform-feats --utt2spk=ark:data/train/split30/1/utt2spk ark:exp/tri3_ali/trans.1 ark:- ark:- | had nonzero return status 36096

feat-to-dim 'ark,s,cs:utils/subset_scp.pl --quiet 333 data/train/split30/1/feats.scp | apply-cmvn  --utt2spk=ark:data/train/split30/1/utt2spk scp:data/train/split30/1/cmvn.scp scp:- ark:- | splice-feats --left-context=3 --right-context=3 ark:- ark:- | transform-feats exp/tri4_nnet/final.mat ark:- ark:- | transform-feats --utt2spk=ark:data/train/split30/1/utt2spk ark:exp/tri3_ali/trans.1 ark:- ark:- | splice-feats --left-context=4 --right-context=4 ark:- ark:- |' -

apply-cmvn --utt2spk=ark:data/train/split30/1/utt2spk scp:data/train/split30/1/cmvn.scp scp:- ark:-

splice-feats --left-context=4 --right-context=4 ark:- ark:-

transform-feats exp/tri4_nnet/final.mat ark:- ark:-

splice-feats --left-context=3 --right-context=3 ark:- ark:-

transform-feats --utt2spk=ark:data/train/split30/1/utt2spk ark:exp/tri3_ali/trans.1 ark:- ark:-

WARNING (feat-to-dim[5.2.124~1396-70748]:Close():kaldi-io.cc:501) Pipe utils/subset_scp.pl --quiet 333 data/train/split30/1/feats.scp | apply-cmvn  --utt2spk=ark:data/train/split30/1/utt2spk scp:data/train/split30/1/cmvn.scp scp:- ark:- | splice-feats --left-context=3 --right-context=3 ark:- ark:- | transform-feats exp/tri4_nnet/final.mat ark:- ark:- | transform-feats --utt2spk=ark:data/train/split30/1/utt2spk ark:exp/tri3_ali/trans.1 ark:- ark:- | splice-feats --left-context=4 --right-context=4 ark:- ark:- | had nonzero return status 36096

steps/nnet2/get_lda.sh: Accumulating LDA statistics.

steps/nnet2/get_lda.sh: Finished estimating LDA

steps/nnet2/train_tanh.sh: calling get_egs.sh

steps/nnet2/get_egs.sh --transform-dir exp/tri3_ali --splice-width 4 --samples-per-iter 200000 --num-jobs-nnet 30 --stage 0 --cmd run.pl --mem 4G --io-opts --max-jobs-run 5 data/train data/lang exp/tri3_ali exp/tri4_nnet

steps/nnet2/get_egs.sh: feature type is lda

steps/nnet2/get_egs.sh: using transforms from exp/tri3_ali

steps/nnet2/get_egs.sh: working out number of frames of training data

utils/data/get_utt2dur.sh: segments file does not exist so getting durations from wave files

utils/data/get_utt2dur.sh: successfully obtained utterance lengths from sphere-file headers

utils/data/get_utt2dur.sh: computed data/train/utt2dur

feat-to-len 'scp:head -n 10 data/train/feats.scp|' ark,t:-

steps/nnet2/get_egs.sh: Every epoch, splitting the data up into 1 iterations,

steps/nnet2/get_egs.sh: giving samples-per-iteration of 37740 (you requested 200000).

Getting validation and training subset examples.

steps/nnet2/get_egs.sh: extracting validation and training-subset alignments.

copy-int-vector ark:- ark,t:-

LOG (copy-int-vector[5.2.124~1396-70748]:main():copy-int-vector.cc:83) Copied 3696 vectors of int32.

Getting subsets of validation examples for diagnostics and combination.

Creating training examples

Generating training examples on disk

steps/nnet2/get_egs.sh: rearranging examples into parts for different parallel jobs

steps/nnet2/get_egs.sh: Since iters-per-epoch == 1, just concatenating the data.

Shuffling the order of training examples

(in order to avoid stressing the disk, these won't all run at once).

steps/nnet2/get_egs.sh: Finished preparing training examples

steps/nnet2/train_tanh.sh: initializing neural net

Training transition probabilities and setting priors

steps/nnet2/train_tanh.sh: Will train for 15 + 5 epochs, equalling

steps/nnet2/train_tanh.sh: 15 + 5 = 20 iterations,

steps/nnet2/train_tanh.sh: (while reducing learning rate) + (with constant learning rate).

Training neural net (pass 0)

Training neural net (pass 1)

Training neural net (pass 2)

Training neural net (pass 3)

Training neural net (pass 4)

Training neural net (pass 5)

Training neural net (pass 6)

Training neural net (pass 7)

Training neural net (pass 8)

^[[CTraining neural net (pass 9)

Training neural net (pass 10)

Training neural net (pass 11)

Training neural net (pass 12)

Mixing up from 1920 to 5000 components

Training neural net (pass 13)

Training neural net (pass 14)

Training neural net (pass 15)

Training neural net (pass 16)

Training neural net (pass 17)

Training neural net (pass 18)

Training neural net (pass 19)

Setting num_iters_final=5

Getting average posterior for purposes of adjusting the priors.

Re-adjusting priors based on computed posteriors

Done

Cleaning up data

steps/nnet2/remove_egs.sh: Finished deleting examples in exp/tri4_nnet/egs

Removing most of the models

steps/nnet2/decode.sh --cmd run.pl --mem 4G --nj 5 --num-threads 6 --transform-dir exp/tri3/decode_dev exp/tri3/graph data/dev exp/tri4_nnet/decode_dev

steps/nnet2/decode.sh: feature type is lda

steps/nnet2/decode.sh: using transforms from exp/tri3/decode_dev

steps/diagnostic/analyze_lats.sh --cmd run.pl --mem 4G --iter final exp/tri3/graph exp/tri4_nnet/decode_dev

steps/diagnostic/analyze_lats.sh: see stats in exp/tri4_nnet/decode_dev/log/analyze_alignments.log

Overall, lattice depth (10,50,90-percentile)=(7,33,166) and mean=74.3

steps/diagnostic/analyze_lats.sh: see stats in exp/tri4_nnet/decode_dev/log/analyze_lattice_depth_stats.log

score best paths

score confidence and timing with sclite

Decoding done.

steps/nnet2/decode.sh --cmd run.pl --mem 4G --nj 5 --num-threads 6 --transform-dir exp/tri3/decode_test exp/tri3/graph data/test exp/tri4_nnet/decode_test

steps/nnet2/decode.sh: feature type is lda

steps/nnet2/decode.sh: using transforms from exp/tri3/decode_test

steps/diagnostic/analyze_lats.sh --cmd run.pl --mem 4G --iter final exp/tri3/graph exp/tri4_nnet/decode_test

steps/diagnostic/analyze_lats.sh: see stats in exp/tri4_nnet/decode_test/log/analyze_alignments.log

Overall, lattice depth (10,50,90-percentile)=(7,37,190) and mean=89.1

steps/diagnostic/analyze_lats.sh: see stats in exp/tri4_nnet/decode_test/log/analyze_lattice_depth_stats.log

score best paths

score confidence and timing with sclite

Decoding done.

============================================================================

                    System Combination (DNN+SGMM)

============================================================================

============================================================================

               DNN Hybrid Training & Decoding (Karel's recipe)

============================================================================

steps/nnet/make_fmllr_feats.sh --nj 10 --cmd run.pl --mem 4G --transform-dir exp/tri3/decode_test data-fmllr-tri3/test data/test exp/tri3 data-fmllr-tri3/test/log data-fmllr-tri3/test/data

steps/nnet/make_fmllr_feats.sh: feature type is lda_fmllr

utils/copy_data_dir.sh: copied data from data/test to data-fmllr-tri3/test

utils/validate_data_dir.sh: Successfully validated data-directory data-fmllr-tri3/test

steps/nnet/make_fmllr_feats.sh: Done!, type lda_fmllr, data/test --> data-fmllr-tri3/test, using : raw-trans None, gmm exp/tri3, trans exp/tri3/decode_test

steps/nnet/make_fmllr_feats.sh --nj 10 --cmd run.pl --mem 4G --transform-dir exp/tri3/decode_dev data-fmllr-tri3/dev data/dev exp/tri3 data-fmllr-tri3/dev/log data-fmllr-tri3/dev/data

steps/nnet/make_fmllr_feats.sh: feature type is lda_fmllr

utils/copy_data_dir.sh: copied data from data/dev to data-fmllr-tri3/dev

utils/validate_data_dir.sh: Successfully validated data-directory data-fmllr-tri3/dev

steps/nnet/make_fmllr_feats.sh: Done!, type lda_fmllr, data/dev --> data-fmllr-tri3/dev, using : raw-trans None, gmm exp/tri3, trans exp/tri3/decode_dev

steps/nnet/make_fmllr_feats.sh --nj 10 --cmd run.pl --mem 4G --transform-dir exp/tri3_ali data-fmllr-tri3/train data/train exp/tri3 data-fmllr-tri3/train/log data-fmllr-tri3/train/data

steps/nnet/make_fmllr_feats.sh: feature type is lda_fmllr

utils/copy_data_dir.sh: copied data from data/train to data-fmllr-tri3/train

utils/validate_data_dir.sh: Successfully validated data-directory data-fmllr-tri3/train

steps/nnet/make_fmllr_feats.sh: Done!, type lda_fmllr, data/train --> data-fmllr-tri3/train, using : raw-trans None, gmm exp/tri3, trans exp/tri3_ali

utils/subset_data_dir_tr_cv.sh data-fmllr-tri3/train data-fmllr-tri3/train_tr90 data-fmllr-tri3/train_cv10

/home/wenba/source_code/kaldi-trunk/egs/timit/s5/utils/subset_data_dir.sh: reducing #utt from 3696 to 3320

/home/wenba/source_code/kaldi-trunk/egs/timit/s5/utils/subset_data_dir.sh: reducing #utt from 3696 to 376

# steps/nnet/pretrain_dbn.sh --hid-dim 1024 --rbm-iter 20 data-fmllr-tri3/train exp/dnn4_pretrain-dbn

# Started at Fri Sep 15 16:21:28 CST 2017

#

steps/nnet/pretrain_dbn.sh --hid-dim 1024 --rbm-iter 20 data-fmllr-tri3/train exp/dnn4_pretrain-dbn

# INFO

steps/nnet/pretrain_dbn.sh : Pre-training Deep Belief Network as a stack of RBMs

     dir       : exp/dnn4_pretrain-dbn

     Train-set : data-fmllr-tri3/train ''

LOG ([5.2.124~1396-70748]:main():cuda-gpu-available.cc:49) 

### IS CUDA GPU AVAILABLE? 'welen-pc' ###

WARNING ([5.2.124~1396-70748]:SelectGpuId():cu-device.cc:182) Suggestion: use 'nvidia-smi -c 3' to set compute exclusive mode

LOG ([5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:300) Selecting from 1 GPUs

LOG ([5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:315) cudaSetDevice(0): GeForce GTX 980 Ti    free:5631M, used:444M, total:6076M, free/total:0.926827

LOG ([5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:364) Trying to select device: 0 (automatically), mem_ratio: 0.926827

LOG ([5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:383) Success selecting device 0 free mem ratio: 0.926827

LOG ([5.2.124~1396-70748]:FinalizeActiveGpu():cu-device.cc:225) The active GPU is [0]: GeForce GTX 980 Ti    free:5601M, used:474M, total:6076M, free/total:0.921889 version 5.2

### HURRAY, WE GOT A CUDA GPU FOR COMPUTATION!!! ##

### Testing CUDA setup with a small computation (setup = cuda-toolkit + gpu-driver + kaldi):

### Test OK!

# PREPARING FEATURES

copy-feats --compress=true scp:data-fmllr-tri3/train/feats.scp ark,scp:/tmp/kaldi.0Hlp/train.ark,exp/dnn4_pretrain-dbn/train_sorted.scp

LOG (copy-feats[5.2.124~1396-70748]:main():copy-feats.cc:143) Copied 3696 feature matrices.

# 'apply-cmvn' not used,

feat-to-dim 'ark:copy-feats scp:exp/dnn4_pretrain-dbn/train.scp ark:- |' -

copy-feats scp:exp/dnn4_pretrain-dbn/train.scp ark:-

WARNING (feat-to-dim[5.2.124~1396-70748]:Close():kaldi-io.cc:501) Pipe copy-feats scp:exp/dnn4_pretrain-dbn/train.scp ark:- | had nonzero return status 36096

# feature dim : 40 (input of 'feature_transform')

+ default 'feature_transform_proto' with splice +/-5 frames

nnet-initialize --binary=false exp/dnn4_pretrain-dbn/splice5.proto exp/dnn4_pretrain-dbn/tr_splice5.nnet

VLOG[1] (nnet-initialize[5.2.124~1396-70748]:Init():nnet-nnet.cc:314) <Splice> <InputDim> 40 <OutputDim> 440 <BuildVector> -5:5 </BuildVector>

LOG (nnet-initialize[5.2.124~1396-70748]:main():nnet-initialize.cc:63) Written initialized model to exp/dnn4_pretrain-dbn/tr_splice5.nnet

# compute normalization stats from 10k sentences

compute-cmvn-stats ark:- exp/dnn4_pretrain-dbn/cmvn-g.stats

nnet-forward --print-args=true --use-gpu=yes exp/dnn4_pretrain-dbn/tr_splice5.nnet 'ark:copy-feats scp:exp/dnn4_pretrain-dbn/train.scp.10k ark:- |' ark:-

WARNING (nnet-forward[5.2.124~1396-70748]:SelectGpuId():cu-device.cc:182) Suggestion: use 'nvidia-smi -c 3' to set compute exclusive mode

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:300) Selecting from 1 GPUs

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:315) cudaSetDevice(0): GeForce GTX 980 Ti    free:5631M, used:444M, total:6076M, free/total:0.926827

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:364) Trying to select device: 0 (automatically), mem_ratio: 0.926827

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:383) Success selecting device 0 free mem ratio: 0.926827

LOG (nnet-forward[5.2.124~1396-70748]:FinalizeActiveGpu():cu-device.cc:225) The active GPU is [0]: GeForce GTX 980 Ti    free:5601M, used:474M, total:6076M, free/total:0.921889 version 5.2

copy-feats scp:exp/dnn4_pretrain-dbn/train.scp.10k ark:-

LOG (copy-feats[5.2.124~1396-70748]:main():copy-feats.cc:143) Copied 3696 feature matrices.

LOG (nnet-forward[5.2.124~1396-70748]:main():nnet-forward.cc:192) Done 3696 files in 0.169012min, (fps 110921)

LOG (compute-cmvn-stats[5.2.124~1396-70748]:main():compute-cmvn-stats.cc:168) Wrote global CMVN stats to exp/dnn4_pretrain-dbn/cmvn-g.stats

LOG (compute-cmvn-stats[5.2.124~1396-70748]:main():compute-cmvn-stats.cc:171) Done accumulating CMVN stats for 3696 utterances; 0 had errors.

# + normalization of NN-input at 'exp/dnn4_pretrain-dbn/tr_splice5_cmvn-g.nnet'

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:53) Reading exp/dnn4_pretrain-dbn/tr_splice5.nnet

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:65) Concatenating cmvn-to-nnet exp/dnn4_pretrain-dbn/cmvn-g.stats -|

cmvn-to-nnet exp/dnn4_pretrain-dbn/cmvn-g.stats -

LOG (cmvn-to-nnet[5.2.124~1396-70748]:main():cmvn-to-nnet.cc:114) Written cmvn in 'nnet1' model to: -

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:82) Written model to exp/dnn4_pretrain-dbn/tr_splice5_cmvn-g.nnet

### Showing the final 'feature_transform':

nnet-info exp/dnn4_pretrain-dbn/tr_splice5_cmvn-g.nnet

LOG (nnet-info[5.2.124~1396-70748]:main():nnet-info.cc:57) Printed info about exp/dnn4_pretrain-dbn/tr_splice5_cmvn-g.nnet

num-components 3

input-dim 40

output-dim 440

number-of-parameters 0.00088 millions

component 1 : <Splice>, input-dim 40, output-dim 440,

  frame_offsets [ -5 -4 -3 -2 -1 0 1 2 3 4 5 ]

component 2 : <AddShift>, input-dim 440, output-dim 440,

  shift_data ( min -0.190971, max 0.0653268, mean -0.00386261, stddev 0.0387812, skewness -2.779, kurtosis 9.50268 ) , lr-coef 0

component 3 : <Rescale>, input-dim 440, output-dim 440,

  scale_data ( min 0.318242, max 0.975994, mean 0.761088, stddev 0.156292, skewness -0.679222, kurtosis -0.139114 ) , lr-coef 0

###

# PRE-TRAINING RBM LAYER 1

# initializing 'exp/dnn4_pretrain-dbn/1.rbm.init'

# pretraining 'exp/dnn4_pretrain-dbn/1.rbm' (input gauss, lrate 0.01, iters 40)

# converting RBM to exp/dnn4_pretrain-dbn/1.dbn

rbm-convert-to-nnet exp/dnn4_pretrain-dbn/1.rbm exp/dnn4_pretrain-dbn/1.dbn

LOG (rbm-convert-to-nnet[5.2.124~1396-70748]:main():rbm-convert-to-nnet.cc:69) Written model to exp/dnn4_pretrain-dbn/1.dbn

# PRE-TRAINING RBM LAYER 2

# computing cmvn stats 'exp/dnn4_pretrain-dbn/2.cmvn' for RBM initialization

WARNING (nnet-forward[5.2.124~1396-70748]:SelectGpuId():cu-device.cc:182) Suggestion: use 'nvidia-smi -c 3' to set compute exclusive mode

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:300) Selecting from 1 GPUs

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:315) cudaSetDevice(0): GeForce GTX 980 Ti    free:5632M, used:443M, total:6076M, free/total:0.92697

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:364) Trying to select device: 0 (automatically), mem_ratio: 0.92697

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:383) Success selecting device 0 free mem ratio: 0.92697

LOG (nnet-forward[5.2.124~1396-70748]:FinalizeActiveGpu():cu-device.cc:225) The active GPU is [0]: GeForce GTX 980 Ti    free:5602M, used:473M, total:6076M, free/total:0.922033 version 5.2

nnet-concat exp/dnn4_pretrain-dbn/final.feature_transform exp/dnn4_pretrain-dbn/1.dbn -

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:53) Reading exp/dnn4_pretrain-dbn/final.feature_transform

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:65) Concatenating exp/dnn4_pretrain-dbn/1.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:82) Written model to -

copy-feats scp:exp/dnn4_pretrain-dbn/train.scp.10k ark:-

LOG (copy-feats[5.2.124~1396-70748]:main():copy-feats.cc:143) Copied 3696 feature matrices.

LOG (nnet-forward[5.2.124~1396-70748]:main():nnet-forward.cc:192) Done 3696 files in 0.359421min, (fps 52159)

LOG (compute-cmvn-stats[5.2.124~1396-70748]:main():compute-cmvn-stats.cc:168) Wrote global CMVN stats to standard output

LOG (compute-cmvn-stats[5.2.124~1396-70748]:main():compute-cmvn-stats.cc:171) Done accumulating CMVN stats for 3696 utterances; 0 had errors.

LOG (cmvn-to-nnet[5.2.124~1396-70748]:main():cmvn-to-nnet.cc:114) Written cmvn in 'nnet1' model to: exp/dnn4_pretrain-dbn/2.cmvn

initializing 'exp/dnn4_pretrain-dbn/2.rbm.init'

pretraining 'exp/dnn4_pretrain-dbn/2.rbm' (lrate 0.4, iters 20)

# appending RBM to exp/dnn4_pretrain-dbn/2.dbn

nnet-concat exp/dnn4_pretrain-dbn/1.dbn 'rbm-convert-to-nnet exp/dnn4_pretrain-dbn/2.rbm - |' exp/dnn4_pretrain-dbn/2.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:53) Reading exp/dnn4_pretrain-dbn/1.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:65) Concatenating rbm-convert-to-nnet exp/dnn4_pretrain-dbn/2.rbm - |

rbm-convert-to-nnet exp/dnn4_pretrain-dbn/2.rbm -

LOG (rbm-convert-to-nnet[5.2.124~1396-70748]:main():rbm-convert-to-nnet.cc:69) Written model to -

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:82) Written model to exp/dnn4_pretrain-dbn/2.dbn

# PRE-TRAINING RBM LAYER 3

# computing cmvn stats 'exp/dnn4_pretrain-dbn/3.cmvn' for RBM initialization

WARNING (nnet-forward[5.2.124~1396-70748]:SelectGpuId():cu-device.cc:182) Suggestion: use 'nvidia-smi -c 3' to set compute exclusive mode

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:300) Selecting from 1 GPUs

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:315) cudaSetDevice(0): GeForce GTX 980 Ti    free:5632M, used:443M, total:6076M, free/total:0.92697

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:364) Trying to select device: 0 (automatically), mem_ratio: 0.92697

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:383) Success selecting device 0 free mem ratio: 0.92697

LOG (nnet-forward[5.2.124~1396-70748]:FinalizeActiveGpu():cu-device.cc:225) The active GPU is [0]: GeForce GTX 980 Ti    free:5602M, used:473M, total:6076M, free/total:0.922033 version 5.2

nnet-concat exp/dnn4_pretrain-dbn/final.feature_transform exp/dnn4_pretrain-dbn/2.dbn -

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:53) Reading exp/dnn4_pretrain-dbn/final.feature_transform

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:65) Concatenating exp/dnn4_pretrain-dbn/2.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:82) Written model to -

copy-feats scp:exp/dnn4_pretrain-dbn/train.scp.10k ark:-

LOG (copy-feats[5.2.124~1396-70748]:main():copy-feats.cc:143) Copied 3696 feature matrices.

LOG (nnet-forward[5.2.124~1396-70748]:main():nnet-forward.cc:192) Done 3696 files in 0.35735min, (fps 52461.3)

LOG (compute-cmvn-stats[5.2.124~1396-70748]:main():compute-cmvn-stats.cc:168) Wrote global CMVN stats to standard output

LOG (compute-cmvn-stats[5.2.124~1396-70748]:main():compute-cmvn-stats.cc:171) Done accumulating CMVN stats for 3696 utterances; 0 had errors.

LOG (cmvn-to-nnet[5.2.124~1396-70748]:main():cmvn-to-nnet.cc:114) Written cmvn in 'nnet1' model to: exp/dnn4_pretrain-dbn/3.cmvn

initializing 'exp/dnn4_pretrain-dbn/3.rbm.init'

pretraining 'exp/dnn4_pretrain-dbn/3.rbm' (lrate 0.4, iters 20)

# appending RBM to exp/dnn4_pretrain-dbn/3.dbn

nnet-concat exp/dnn4_pretrain-dbn/2.dbn 'rbm-convert-to-nnet exp/dnn4_pretrain-dbn/3.rbm - |' exp/dnn4_pretrain-dbn/3.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:53) Reading exp/dnn4_pretrain-dbn/2.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:65) Concatenating rbm-convert-to-nnet exp/dnn4_pretrain-dbn/3.rbm - |

rbm-convert-to-nnet exp/dnn4_pretrain-dbn/3.rbm -

LOG (rbm-convert-to-nnet[5.2.124~1396-70748]:main():rbm-convert-to-nnet.cc:69) Written model to -

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:82) Written model to exp/dnn4_pretrain-dbn/3.dbn

# PRE-TRAINING RBM LAYER 4

# computing cmvn stats 'exp/dnn4_pretrain-dbn/4.cmvn' for RBM initialization

WARNING (nnet-forward[5.2.124~1396-70748]:SelectGpuId():cu-device.cc:182) Suggestion: use 'nvidia-smi -c 3' to set compute exclusive mode

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:300) Selecting from 1 GPUs

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:315) cudaSetDevice(0): GeForce GTX 980 Ti    free:5602M, used:474M, total:6076M, free/total:0.921992

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:364) Trying to select device: 0 (automatically), mem_ratio: 0.921992

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:383) Success selecting device 0 free mem ratio: 0.921992

LOG (nnet-forward[5.2.124~1396-70748]:FinalizeActiveGpu():cu-device.cc:225) The active GPU is [0]: GeForce GTX 980 Ti    free:5572M, used:504M, total:6076M, free/total:0.917055 version 5.2

nnet-concat exp/dnn4_pretrain-dbn/final.feature_transform exp/dnn4_pretrain-dbn/3.dbn -

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:53) Reading exp/dnn4_pretrain-dbn/final.feature_transform

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:65) Concatenating exp/dnn4_pretrain-dbn/3.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:82) Written model to -

copy-feats scp:exp/dnn4_pretrain-dbn/train.scp.10k ark:-

LOG (copy-feats[5.2.124~1396-70748]:main():copy-feats.cc:143) Copied 3696 feature matrices.

LOG (nnet-forward[5.2.124~1396-70748]:main():nnet-forward.cc:192) Done 3696 files in 0.435866min, (fps 43011)

LOG (compute-cmvn-stats[5.2.124~1396-70748]:main():compute-cmvn-stats.cc:168) Wrote global CMVN stats to standard output

LOG (compute-cmvn-stats[5.2.124~1396-70748]:main():compute-cmvn-stats.cc:171) Done accumulating CMVN stats for 3696 utterances; 0 had errors.

LOG (cmvn-to-nnet[5.2.124~1396-70748]:main():cmvn-to-nnet.cc:114) Written cmvn in 'nnet1' model to: exp/dnn4_pretrain-dbn/4.cmvn

initializing 'exp/dnn4_pretrain-dbn/4.rbm.init'

pretraining 'exp/dnn4_pretrain-dbn/4.rbm' (lrate 0.4, iters 20)

# appending RBM to exp/dnn4_pretrain-dbn/4.dbn

nnet-concat exp/dnn4_pretrain-dbn/3.dbn 'rbm-convert-to-nnet exp/dnn4_pretrain-dbn/4.rbm - |' exp/dnn4_pretrain-dbn/4.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:53) Reading exp/dnn4_pretrain-dbn/3.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:65) Concatenating rbm-convert-to-nnet exp/dnn4_pretrain-dbn/4.rbm - |

rbm-convert-to-nnet exp/dnn4_pretrain-dbn/4.rbm -

LOG (rbm-convert-to-nnet[5.2.124~1396-70748]:main():rbm-convert-to-nnet.cc:69) Written model to -

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:82) Written model to exp/dnn4_pretrain-dbn/4.dbn

# PRE-TRAINING RBM LAYER 5

# computing cmvn stats 'exp/dnn4_pretrain-dbn/5.cmvn' for RBM initialization

WARNING (nnet-forward[5.2.124~1396-70748]:SelectGpuId():cu-device.cc:182) Suggestion: use 'nvidia-smi -c 3' to set compute exclusive mode

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:300) Selecting from 1 GPUs

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:315) cudaSetDevice(0): GeForce GTX 980 Ti    free:5578M, used:498M, total:6076M, free/total:0.918022

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:364) Trying to select device: 0 (automatically), mem_ratio: 0.918022

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:383) Success selecting device 0 free mem ratio: 0.918022

LOG (nnet-forward[5.2.124~1396-70748]:FinalizeActiveGpu():cu-device.cc:225) The active GPU is [0]: GeForce GTX 980 Ti    free:5548M, used:528M, total:6076M, free/total:0.913085 version 5.2

nnet-concat exp/dnn4_pretrain-dbn/final.feature_transform exp/dnn4_pretrain-dbn/4.dbn -

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:53) Reading exp/dnn4_pretrain-dbn/final.feature_transform

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:65) Concatenating exp/dnn4_pretrain-dbn/4.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:82) Written model to -

copy-feats scp:exp/dnn4_pretrain-dbn/train.scp.10k ark:-

LOG (copy-feats[5.2.124~1396-70748]:main():copy-feats.cc:143) Copied 3696 feature matrices.

LOG (nnet-forward[5.2.124~1396-70748]:main():nnet-forward.cc:192) Done 3696 files in 0.364739min, (fps 51398.6)

LOG (compute-cmvn-stats[5.2.124~1396-70748]:main():compute-cmvn-stats.cc:168) Wrote global CMVN stats to standard output

LOG (compute-cmvn-stats[5.2.124~1396-70748]:main():compute-cmvn-stats.cc:171) Done accumulating CMVN stats for 3696 utterances; 0 had errors.

LOG (cmvn-to-nnet[5.2.124~1396-70748]:main():cmvn-to-nnet.cc:114) Written cmvn in 'nnet1' model to: exp/dnn4_pretrain-dbn/5.cmvn

initializing 'exp/dnn4_pretrain-dbn/5.rbm.init'

pretraining 'exp/dnn4_pretrain-dbn/5.rbm' (lrate 0.4, iters 20)

# appending RBM to exp/dnn4_pretrain-dbn/5.dbn

nnet-concat exp/dnn4_pretrain-dbn/4.dbn 'rbm-convert-to-nnet exp/dnn4_pretrain-dbn/5.rbm - |' exp/dnn4_pretrain-dbn/5.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:53) Reading exp/dnn4_pretrain-dbn/4.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:65) Concatenating rbm-convert-to-nnet exp/dnn4_pretrain-dbn/5.rbm - |

rbm-convert-to-nnet exp/dnn4_pretrain-dbn/5.rbm -

LOG (rbm-convert-to-nnet[5.2.124~1396-70748]:main():rbm-convert-to-nnet.cc:69) Written model to -

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:82) Written model to exp/dnn4_pretrain-dbn/5.dbn

# PRE-TRAINING RBM LAYER 6

# computing cmvn stats 'exp/dnn4_pretrain-dbn/6.cmvn' for RBM initialization

WARNING (nnet-forward[5.2.124~1396-70748]:SelectGpuId():cu-device.cc:182) Suggestion: use 'nvidia-smi -c 3' to set compute exclusive mode

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:300) Selecting from 1 GPUs

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:315) cudaSetDevice(0): GeForce GTX 980 Ti    free:5605M, used:470M, total:6076M, free/total:0.922527

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:364) Trying to select device: 0 (automatically), mem_ratio: 0.922527

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:383) Success selecting device 0 free mem ratio: 0.922527

LOG (nnet-forward[5.2.124~1396-70748]:FinalizeActiveGpu():cu-device.cc:225) The active GPU is [0]: GeForce GTX 980 Ti    free:5575M, used:500M, total:6076M, free/total:0.91759 version 5.2

nnet-concat exp/dnn4_pretrain-dbn/final.feature_transform exp/dnn4_pretrain-dbn/5.dbn -

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:53) Reading exp/dnn4_pretrain-dbn/final.feature_transform

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:65) Concatenating exp/dnn4_pretrain-dbn/5.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:82) Written model to -

copy-feats scp:exp/dnn4_pretrain-dbn/train.scp.10k ark:-

LOG (copy-feats[5.2.124~1396-70748]:main():copy-feats.cc:143) Copied 3696 feature matrices.

LOG (nnet-forward[5.2.124~1396-70748]:main():nnet-forward.cc:192) Done 3696 files in 0.403902min, (fps 46414.9)

LOG (compute-cmvn-stats[5.2.124~1396-70748]:main():compute-cmvn-stats.cc:168) Wrote global CMVN stats to standard output

LOG (compute-cmvn-stats[5.2.124~1396-70748]:main():compute-cmvn-stats.cc:171) Done accumulating CMVN stats for 3696 utterances; 0 had errors.

LOG (cmvn-to-nnet[5.2.124~1396-70748]:main():cmvn-to-nnet.cc:114) Written cmvn in 'nnet1' model to: exp/dnn4_pretrain-dbn/6.cmvn

initializing 'exp/dnn4_pretrain-dbn/6.rbm.init'

pretraining 'exp/dnn4_pretrain-dbn/6.rbm' (lrate 0.4, iters 20)

# appending RBM to exp/dnn4_pretrain-dbn/6.dbn

nnet-concat exp/dnn4_pretrain-dbn/5.dbn 'rbm-convert-to-nnet exp/dnn4_pretrain-dbn/6.rbm - |' exp/dnn4_pretrain-dbn/6.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:53) Reading exp/dnn4_pretrain-dbn/5.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:65) Concatenating rbm-convert-to-nnet exp/dnn4_pretrain-dbn/6.rbm - |

rbm-convert-to-nnet exp/dnn4_pretrain-dbn/6.rbm -

LOG (rbm-convert-to-nnet[5.2.124~1396-70748]:main():rbm-convert-to-nnet.cc:69) Written model to -

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:82) Written model to exp/dnn4_pretrain-dbn/6.dbn

# REPORT

# RBM pre-training progress (line per-layer)

exp/dnn4_pretrain-dbn/log/rbm.1.log:progress: [69.2148 60.1847 57.6252 55.9637 54.9476 54.1765 53.5664 53.1591 52.8667 52.6254 52.3768 52.223 51.9968 51.8111 51.7657 51.5485 51.4773 51.4296 51.2532 51.2139 51.1692 51.0821 51.0227 50.9377 50.9296 50.898 50.8085 50.8238 50.7534 50.7348 50.7076 50.6632 50.6615 50.6518 50.6302 50.5964 50.5804 50.5564 50.513 50.5592 50.4797 50.4993 50.51 50.4201 50.4663 50.4791 50.4227 50.4098 50.3793 50.4227 50.389 50.3265 50.4077 50.3461 50.3283 50.3104 50.3 50.3348 50.3243 50.34 50.2997 50.3534 50.3074 50.2881 50.3791 50.3117 50.334 50.3648 50.2924 50.3353 50.3746 50.3269 50.344 50.3242 50.3575 50.3607 50.3162 50.3958 50.321 50.3486 50.3444 50.3016 50.346 50.3465 50.3456 50.3346 50.3631 50.322 50.2912 50.355 50.3015 50.3125 50.3776 50.2883 50.3309 50.366 50.3244 50.3295 50.3156 50.357 50.3315 50.2989 50.3498 50.3214 50.2986 50.326 50.2989 50.3249 50.3351 50.33 50.3258 50.3376 50.2808 50.2987 50.3308 50.2853 50.2821 50.3463 50.2576 50.3191 50.3351 50.2726 50.311 50.2955 ]

exp/dnn4_pretrain-dbn/log/rbm.2.log:progress: [9.47593 6.68198 5.9925 5.79754 5.69381 5.61983 5.57938 5.55269 5.51679 5.50825 5.48539 5.46428 5.45513 5.44116 5.42658 5.40876 5.40853 5.39329 5.37245 5.37511 5.3633 5.3483 5.34547 5.3275 5.32134 5.31471 5.29894 5.29564 5.28453 5.27386 5.25925 5.25488 5.24767 5.23009 5.22815 5.2176 5.20372 5.19856 5.19204 5.18034 5.17081 5.16935 5.15679 5.14222 5.14116 5.13264 5.11946 5.11546 5.10223 5.0995 5.09251 5.08136 5.07479 5.06829 5.05639 5.05179 5.04034 5.04904 5.04462 5.05154 5.04966 5.04821 ]

exp/dnn4_pretrain-dbn/log/rbm.3.log:progress: [8.82161 5.93724 5.17477 4.88171 4.72492 4.62911 4.57975 4.54986 4.5189 4.51286 4.49688 4.47944 4.4744 4.46463 4.4546 4.44478 4.44225 4.4341 4.42407 4.42035 4.41003 4.40472 4.39895 4.38925 4.38713 4.38257 4.37139 4.36597 4.35891 4.35509 4.34409 4.33953 4.3354 4.32333 4.32584 4.3164 4.30591 4.30421 4.29929 4.2915 4.28396 4.28321 4.27594 4.26477 4.26634 4.2582 4.25073 4.24657 4.23891 4.23711 4.23202 4.22429 4.22139 4.22028 4.21185 4.20091 4.20112 4.20474 4.19747 4.20566 4.20243 4.20412 ]

exp/dnn4_pretrain-dbn/log/rbm.4.log:progress: [6.47138 4.39349 3.95403 3.77159 3.67085 3.60859 3.57098 3.54628 3.51907 3.51778 3.50106 3.48814 3.48639 3.47567 3.46829 3.46467 3.4607 3.45052 3.4467 3.44301 3.44124 3.43569 3.42892 3.4242 3.42692 3.41555 3.4108 3.40813 3.40705 3.40038 3.39258 3.39316 3.38946 3.37972 3.3854 3.37847 3.37083 3.37235 3.36492 3.35975 3.36044 3.35709 3.34989 3.34552 3.34573 3.34155 3.34065 3.33502 3.33037 3.33208 3.32376 3.32243 3.31974 3.31912 3.31325 3.30906 3.30656 3.31356 3.30566 3.31492 3.31235 3.30959 ]

exp/dnn4_pretrain-dbn/log/rbm.5.log:progress: [6.31728 4.20671 3.63834 3.39393 3.24875 3.17134 3.13306 3.10876 3.08828 3.08771 3.07425 3.06279 3.06532 3.0552 3.04786 3.04868 3.04386 3.03695 3.03614 3.03225 3.02967 3.02828 3.01727 3.01428 3.01825 3.01191 3.0072 3.00605 3.00383 2.99691 2.99246 2.99273 2.99128 2.98259 2.98507 2.98083 2.97321 2.97814 2.97144 2.96666 2.96739 2.96531 2.95768 2.95748 2.95505 2.95458 2.95264 2.94641 2.94112 2.94598 2.94086 2.93676 2.93529 2.93577 2.92937 2.92661 2.92618 2.92859 2.92679 2.93277 2.93028 2.92924 ]

exp/dnn4_pretrain-dbn/log/rbm.6.log:progress: [4.69362 3.23743 2.88821 2.72413 2.63688 2.58968 2.55744 2.53789 2.51894 2.51731 2.50398 2.49555 2.49793 2.49135 2.48345 2.48715 2.47985 2.47587 2.47577 2.46986 2.46492 2.46922 2.46131 2.46046 2.46231 2.45514 2.45273 2.45314 2.45089 2.44601 2.44129 2.44196 2.44055 2.43424 2.44036 2.43207 2.42852 2.43271 2.42498 2.42388 2.4243 2.42063 2.41817 2.41862 2.41386 2.4136 2.4153 2.40774 2.40672 2.41052 2.40388 2.40278 2.4025 2.40106 2.39694 2.39437 2.39436 2.39827 2.39453 2.40005 2.3956 2.39306 ]

Pre-training finished.

# Removing features tmpdir /tmp/kaldi.0Hlp @ welen-pc

train.ark

# Accounting: time=2620 threads=1

# Ended (code 0) at Fri Sep 15 17:05:08 CST 2017, elapsed time 2620 seconds

# steps/nnet/train.sh --feature-transform exp/dnn4_pretrain-dbn/final.feature_transform --dbn exp/dnn4_pretrain-dbn/6.dbn --hid-layers 0 --learn-rate 0.008 data-fmllr-tri3/train_tr90 data-fmllr-tri3/train_cv10 data/lang exp/tri3_ali exp/tri3_ali exp/dnn4_pretrain-dbn_dnn

# Started at Fri Sep 15 17:05:08 CST 2017

#

steps/nnet/train.sh --feature-transform exp/dnn4_pretrain-dbn/final.feature_transform --dbn exp/dnn4_pretrain-dbn/6.dbn --hid-layers 0 --learn-rate 0.008 data-fmllr-tri3/train_tr90 data-fmllr-tri3/train_cv10 data/lang exp/tri3_ali exp/tri3_ali exp/dnn4_pretrain-dbn_dnn

# INFO

steps/nnet/train.sh : Training Neural Network

     dir       : exp/dnn4_pretrain-dbn_dnn

     Train-set : data-fmllr-tri3/train_tr90 3320, exp/tri3_ali

     CV-set    : data-fmllr-tri3/train_cv10 376 exp/tri3_ali 

LOG ([5.2.124~1396-70748]:main():cuda-gpu-available.cc:49) 

### IS CUDA GPU AVAILABLE? 'welen-pc' ###

WARNING ([5.2.124~1396-70748]:SelectGpuId():cu-device.cc:182) Suggestion: use 'nvidia-smi -c 3' to set compute exclusive mode

LOG ([5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:300) Selecting from 1 GPUs

LOG ([5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:315) cudaSetDevice(0): GeForce GTX 980 Ti    free:5603M, used:473M, total:6076M, free/total:0.922136

LOG ([5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:364) Trying to select device: 0 (automatically), mem_ratio: 0.922136

LOG ([5.2.124~1396-70748]:SelectGpuIdAuto():cu-device.cc:383) Success selecting device 0 free mem ratio: 0.922136

LOG ([5.2.124~1396-70748]:FinalizeActiveGpu():cu-device.cc:225) The active GPU is [0]: GeForce GTX 980 Ti    free:5573M, used:503M, total:6076M, free/total:0.917199 version 5.2

### HURRAY, WE GOT A CUDA GPU FOR COMPUTATION!!! ##

### Testing CUDA setup with a small computation (setup = cuda-toolkit + gpu-driver + kaldi):

### Test OK!

# PREPARING ALIGNMENTS

Using PDF targets from dirs 'exp/tri3_ali' 'exp/tri3_ali'

hmm-info exp/tri3_ali/final.mdl

copy-transition-model --binary=false exp/tri3_ali/final.mdl exp/dnn4_pretrain-dbn_dnn/final.mdl

LOG (copy-transition-model[5.2.124~1396-70748]:main():copy-transition-model.cc:62) Copied transition model.

# PREPARING FEATURES

# re-saving features to local disk,

copy-feats --compress=true scp:data-fmllr-tri3/train_tr90/feats.scp ark,scp:/tmp/kaldi.B9DQ/train.ark,exp/dnn4_pretrain-dbn_dnn/train_sorted.scp

LOG (copy-feats[5.2.124~1396-70748]:main():copy-feats.cc:143) Copied 3320 feature matrices.

copy-feats --compress=true scp:data-fmllr-tri3/train_cv10/feats.scp ark,scp:/tmp/kaldi.B9DQ/cv.ark,exp/dnn4_pretrain-dbn_dnn/cv.scp

LOG (copy-feats[5.2.124~1396-70748]:main():copy-feats.cc:143) Copied 376 feature matrices.

# importing feature settings from dir 'exp/dnn4_pretrain-dbn'

# cmvn_opts='' delta_opts='' ivector_dim=''

# 'apply-cmvn' is not used,

feat-to-dim 'ark:copy-feats scp:exp/dnn4_pretrain-dbn_dnn/train.scp ark:- |' -

copy-feats scp:exp/dnn4_pretrain-dbn_dnn/train.scp ark:-

WARNING (feat-to-dim[5.2.124~1396-70748]:Close():kaldi-io.cc:501) Pipe copy-feats scp:exp/dnn4_pretrain-dbn_dnn/train.scp ark:- | had nonzero return status 36096

# feature dim : 40 (input of 'feature_transform')

# importing 'feature_transform' from 'exp/dnn4_pretrain-dbn/final.feature_transform'

### Showing the final 'feature_transform':

nnet-info exp/dnn4_pretrain-dbn_dnn/imported_final.feature_transform

LOG (nnet-info[5.2.124~1396-70748]:main():nnet-info.cc:57) Printed info about exp/dnn4_pretrain-dbn_dnn/imported_final.feature_transform

num-components 3

input-dim 40

output-dim 440

number-of-parameters 0.00088 millions

component 1 : <Splice>, input-dim 40, output-dim 440,

  frame_offsets [ -5 -4 -3 -2 -1 0 1 2 3 4 5 ]

component 2 : <AddShift>, input-dim 440, output-dim 440,

  shift_data ( min -0.190971, max 0.0653268, mean -0.00386261, stddev 0.0387812, skewness -2.779, kurtosis 9.50268 ) , lr-coef 0

component 3 : <Rescale>, input-dim 440, output-dim 440,

  scale_data ( min 0.318242, max 0.975994, mean 0.761088, stddev 0.156292, skewness -0.679222, kurtosis -0.139114 ) , lr-coef 0

###

# NN-INITIALIZATION

# getting input/output dims :

feat-to-dim 'ark:copy-feats scp:exp/dnn4_pretrain-dbn_dnn/train.scp ark:- | nnet-forward "nnet-concat exp/dnn4_pretrain-dbn_dnn/final.feature_transform '\''exp/dnn4_pretrain-dbn/6.dbn'\'' -|" ark:- ark:- |' -

copy-feats scp:exp/dnn4_pretrain-dbn_dnn/train.scp ark:-

nnet-forward "nnet-concat exp/dnn4_pretrain-dbn_dnn/final.feature_transform 'exp/dnn4_pretrain-dbn/6.dbn' -|" ark:- ark:-

LOG (nnet-forward[5.2.124~1396-70748]:SelectGpuId():cu-device.cc:110) Manually selected to compute on CPU.

nnet-concat exp/dnn4_pretrain-dbn_dnn/final.feature_transform exp/dnn4_pretrain-dbn/6.dbn -

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:53) Reading exp/dnn4_pretrain-dbn_dnn/final.feature_transform

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:65) Concatenating exp/dnn4_pretrain-dbn/6.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:82) Written model to -

WARNING (feat-to-dim[5.2.124~1396-70748]:Close():kaldi-io.cc:501) Pipe copy-feats scp:exp/dnn4_pretrain-dbn_dnn/train.scp ark:- | nnet-forward "nnet-concat exp/dnn4_pretrain-dbn_dnn/final.feature_transform 'exp/dnn4_pretrain-dbn/6.dbn' -|" ark:- ark:- | had nonzero return status 36096

# genrating network prototype exp/dnn4_pretrain-dbn_dnn/nnet.proto

['utils/nnet/make_nnet_proto.py', '', '', '', '']

# initializing the NN 'exp/dnn4_pretrain-dbn_dnn/nnet.proto' -> 'exp/dnn4_pretrain-dbn_dnn/nnet.init'

nnet-initialize --seed=777 exp/dnn4_pretrain-dbn_dnn/nnet.proto exp/dnn4_pretrain-dbn_dnn/nnet.init

VLOG[1] (nnet-initialize[5.2.124~1396-70748]:Init():nnet-nnet.cc:314) <AffineTransform> <InputDim> 1024 <OutputDim> 1920 <BiasMean> 0.000000 <BiasRange> 0.000000 <ParamStddev> 0.091225

VLOG[1] (nnet-initialize[5.2.124~1396-70748]:Init():nnet-nnet.cc:314) <Softmax> <InputDim> 1920 <OutputDim> 1920

VLOG[1] (nnet-initialize[5.2.124~1396-70748]:Init():nnet-nnet.cc:314) </NnetProto>

LOG (nnet-initialize[5.2.124~1396-70748]:main():nnet-initialize.cc:63) Written initialized model to exp/dnn4_pretrain-dbn_dnn/nnet.init

nnet-concat exp/dnn4_pretrain-dbn/6.dbn exp/dnn4_pretrain-dbn_dnn/nnet.init exp/dnn4_pretrain-dbn_dnn/nnet_dbn_dnn.init

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:53) Reading exp/dnn4_pretrain-dbn/6.dbn

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:65) Concatenating exp/dnn4_pretrain-dbn_dnn/nnet.init

LOG (nnet-concat[5.2.124~1396-70748]:main():nnet-concat.cc:82) Written model to exp/dnn4_pretrain-dbn_dnn/nnet_dbn_dnn.init

# RUNNING THE NN-TRAINING SCHEDULER

steps/nnet/train_scheduler.sh --feature-transform exp/dnn4_pretrain-dbn_dnn/final.feature_transform --learn-rate 0.008 exp/dnn4_pretrain-dbn_dnn/nnet_dbn_dnn.init ark:copy-feats scp:exp/dnn4_pretrain-dbn_dnn/train.scp ark:- | ark:copy-feats scp:exp/dnn4_pretrain-dbn_dnn/cv.scp ark:- | ark:ali-to-pdf exp/tri3_ali/final.mdl "ark:gunzip -c exp/tri3_ali/ali.*.gz |" ark:- | ali-to-post ark:- ark:- | ark:ali-to-pdf exp/tri3_ali/final.mdl "ark:gunzip -c exp/tri3_ali/ali.*.gz |" ark:- | ali-to-post ark:- ark:- | exp/dnn4_pretrain-dbn_dnn

CROSSVAL PRERUN AVG.LOSS 7.7375 (Xent),

ITERATION 01: TRAIN AVG.LOSS 2.1049, (lrate0.008), CROSSVAL AVG.LOSS 1.9152, nnet accepted (nnet_dbn_dnn_iter01_learnrate0.008_tr2.1049_cv1.9152)

ITERATION 02: TRAIN AVG.LOSS 1.4000, (lrate0.008), CROSSVAL AVG.LOSS 1.8016, nnet accepted (nnet_dbn_dnn_iter02_learnrate0.008_tr1.4000_cv1.8016)

ITERATION 03: TRAIN AVG.LOSS 1.1952, (lrate0.008), CROSSVAL AVG.LOSS 1.7656, nnet accepted (nnet_dbn_dnn_iter03_learnrate0.008_tr1.1952_cv1.7656)

ITERATION 04: TRAIN AVG.LOSS 1.0485, (lrate0.008), CROSSVAL AVG.LOSS 1.7654, nnet accepted (nnet_dbn_dnn_iter04_learnrate0.008_tr1.0485_cv1.7654)

ITERATION 05: TRAIN AVG.LOSS 0.8894, (lrate0.004), CROSSVAL AVG.LOSS 1.6597, nnet accepted (nnet_dbn_dnn_iter05_learnrate0.004_tr0.8894_cv1.6597)

ITERATION 06: TRAIN AVG.LOSS 0.8144, (lrate0.002), CROSSVAL AVG.LOSS 1.5910, nnet accepted (nnet_dbn_dnn_iter06_learnrate0.002_tr0.8144_cv1.5910)

ITERATION 07: TRAIN AVG.LOSS 0.7825, (lrate0.001), CROSSVAL AVG.LOSS 1.5467, nnet accepted (nnet_dbn_dnn_iter07_learnrate0.001_tr0.7825_cv1.5467)

ITERATION 08: TRAIN AVG.LOSS 0.7686, (lrate0.0005), CROSSVAL AVG.LOSS 1.5175, nnet accepted (nnet_dbn_dnn_iter08_learnrate0.0005_tr0.7686_cv1.5175)

ITERATION 09: TRAIN AVG.LOSS 0.7614, (lrate0.00025), CROSSVAL AVG.LOSS 1.4999, nnet accepted (nnet_dbn_dnn_iter09_learnrate0.00025_tr0.7614_cv1.4999)

ITERATION 10: TRAIN AVG.LOSS 0.7570, (lrate0.000125), CROSSVAL AVG.LOSS 1.4902, nnet accepted (nnet_dbn_dnn_iter10_learnrate0.000125_tr0.7570_cv1.4902)

ITERATION 11: TRAIN AVG.LOSS 0.7540, (lrate6.25e-05), CROSSVAL AVG.LOSS 1.4857, nnet accepted (nnet_dbn_dnn_iter11_learnrate6.25e-05_tr0.7540_cv1.4857)

ITERATION 12: TRAIN AVG.LOSS 0.7520, (lrate3.125e-05), CROSSVAL AVG.LOSS 1.4839, nnet accepted (nnet_dbn_dnn_iter12_learnrate3.125e-05_tr0.7520_cv1.4839)

ITERATION 13: TRAIN AVG.LOSS 0.7508, (lrate1.5625e-05), CROSSVAL AVG.LOSS 1.4832, nnet accepted (nnet_dbn_dnn_iter13_learnrate1.5625e-05_tr0.7508_cv1.4832)

finished, too small rel. improvement 0.000478475

steps/nnet/train_scheduler.sh: Succeeded training the Neural Network : 'exp/dnn4_pretrain-dbn_dnn/final.nnet'

steps/nnet/train.sh: Successfuly finished. 'exp/dnn4_pretrain-dbn_dnn'

steps/nnet/decode.sh --nj 20 --cmd run.pl --mem 4G --acwt 0.2 exp/tri3/graph data-fmllr-tri3/test exp/dnn4_pretrain-dbn_dnn/decode_test

# Removing features tmpdir /tmp/kaldi.B9DQ @ welen-pc

cv.ark

train.ark

# Accounting: time=375 threads=1

# Ended (code 0) at Fri Sep 15 17:11:23 CST 2017, elapsed time 375 seconds

steps/nnet/decode.sh --nj 20 --cmd run.pl --mem 4G --acwt 0.2 exp/tri3/graph data-fmllr-tri3/dev exp/dnn4_pretrain-dbn_dnn/decode_dev

steps/nnet/align.sh --nj 20 --cmd run.pl --mem 4G data-fmllr-tri3/train data/lang exp/dnn4_pretrain-dbn_dnn exp/dnn4_pretrain-dbn_dnn_ali

steps/nnet/align.sh: aligning data 'data-fmllr-tri3/train' using nnet/model 'exp/dnn4_pretrain-dbn_dnn', putting alignments in 'exp/dnn4_pretrain-dbn_dnn_ali'

steps/nnet/align.sh: done aligning data.

steps/nnet/make_denlats.sh --nj 20 --cmd run.pl --mem 4G --acwt 0.2 --lattice-beam 10.0 --beam 18.0 data-fmllr-tri3/train data/lang exp/dnn4_pretrain-dbn_dnn exp/dnn4_pretrain-dbn_dnn_denlats

Making unigram grammar FST in exp/dnn4_pretrain-dbn_dnn_denlats/lang

Compiling decoding graph in exp/dnn4_pretrain-dbn_dnn_denlats/dengraph

tree-info exp/dnn4_pretrain-dbn_dnn/tree

tree-info exp/dnn4_pretrain-dbn_dnn/tree

fsttablecompose exp/dnn4_pretrain-dbn_dnn_denlats/lang/L_disambig.fst exp/dnn4_pretrain-dbn_dnn_denlats/lang/G.fst

fstminimizeencoded

fstdeterminizestar --use-log=true

fstpushspecial

fstisstochastic exp/dnn4_pretrain-dbn_dnn_denlats/lang/tmp/LG.fst

1.27271e-05 1.27271e-05

fstcomposecontext --context-size=3 --central-position=1 --read-disambig-syms=exp/dnn4_pretrain-dbn_dnn_denlats/lang/phones/disambig.int --write-disambig-syms=exp/dnn4_pretrain-dbn_dnn_denlats/lang/tmp/disambig_ilabels_3_1.int exp/dnn4_pretrain-dbn_dnn_denlats/lang/tmp/ilabels_3_1.19787

fstisstochastic exp/dnn4_pretrain-dbn_dnn_denlats/lang/tmp/CLG_3_1.fst

1.27657e-05 0

make-h-transducer --disambig-syms-out=exp/dnn4_pretrain-dbn_dnn_denlats/dengraph/disambig_tid.int --transition-scale=1.0 exp/dnn4_pretrain-dbn_dnn_denlats/lang/tmp/ilabels_3_1 exp/dnn4_pretrain-dbn_dnn/tree exp/dnn4_pretrain-dbn_dnn/final.mdl

fstdeterminizestar --use-log=true

fsttablecompose exp/dnn4_pretrain-dbn_dnn_denlats/dengraph/Ha.fst exp/dnn4_pretrain-dbn_dnn_denlats/lang/tmp/CLG_3_1.fst

fstrmsymbols exp/dnn4_pretrain-dbn_dnn_denlats/dengraph/disambig_tid.int

fstminimizeencoded

fstrmepslocal

fstisstochastic exp/dnn4_pretrain-dbn_dnn_denlats/dengraph/HCLGa.fst

0.000473932 -0.000484132

add-self-loops --self-loop-scale=0.1 --reorder=true exp/dnn4_pretrain-dbn_dnn/final.mdl

steps/nnet/make_denlats.sh: generating denlats from data 'data-fmllr-tri3/train', putting lattices in 'exp/dnn4_pretrain-dbn_dnn_denlats'

steps/nnet/make_denlats.sh: done generating denominator lattices.

steps/nnet/train_mpe.sh --cmd run.pl --gpu 1 --num-iters 6 --acwt 0.2 --do-smbr true data-fmllr-tri3/train data/lang exp/dnn4_pretrain-dbn_dnn exp/dnn4_pretrain-dbn_dnn_ali exp/dnn4_pretrain-dbn_dnn_denlats exp/dnn4_pretrain-dbn_dnn_smbr

Pass 1 (learnrate 0.00001)

 TRAINING FINISHED; Time taken = 2.88075 min; processed 6507.71 frames per second.

 Done 3696 files, 0 with no reference alignments, 0 with no lattices, 0 with other errors.

 Overall average frame-accuracy is 0.869688 over 1124823 frames.

Pass 2 (learnrate 1e-05)

 TRAINING FINISHED; Time taken = 2.87964 min; processed 6510.22 frames per second.

 Done 3696 files, 0 with no reference alignments, 0 with no lattices, 0 with other errors.

 Overall average frame-accuracy is 0.876663 over 1124823 frames.

Pass 3 (learnrate 1e-05)

 TRAINING FINISHED; Time taken = 2.89735 min; processed 6470.42 frames per second.

 Done 3696 files, 0 with no reference alignments, 0 with no lattices, 0 with other errors.

 Overall average frame-accuracy is 0.880725 over 1124823 frames.

Pass 4 (learnrate 1e-05)

 TRAINING FINISHED; Time taken = 2.85592 min; processed 6564.28 frames per second.

 Done 3696 files, 0 with no reference alignments, 0 with no lattices, 0 with other errors.

 Overall average frame-accuracy is 0.883655 over 1124823 frames.

Pass 5 (learnrate 1e-05)

 TRAINING FINISHED; Time taken = 2.89002 min; processed 6486.83 frames per second.

 Done 3696 files, 0 with no reference alignments, 0 with no lattices, 0 with other errors.

 Overall average frame-accuracy is 0.886018 over 1124823 frames.

Pass 6 (learnrate 1e-05)

 TRAINING FINISHED; Time taken = 2.8828 min; processed 6503.08 frames per second.

 Done 3696 files, 0 with no reference alignments, 0 with no lattices, 0 with other errors.

 Overall average frame-accuracy is 0.888026 over 1124823 frames.

MPE/sMBR training finished

Re-estimating priors by forwarding 10k utterances from training set.

steps/nnet/make_priors.sh --cmd run.pl --mem 4G --nj 20 data-fmllr-tri3/train exp/dnn4_pretrain-dbn_dnn_smbr

Accumulating prior stats by forwarding 'data-fmllr-tri3/train' with 'exp/dnn4_pretrain-dbn_dnn_smbr'

Succeeded creating prior counts 'exp/dnn4_pretrain-dbn_dnn_smbr/prior_counts' from 'data-fmllr-tri3/train'

steps/nnet/train_mpe.sh: Done. 'exp/dnn4_pretrain-dbn_dnn_smbr'

steps/nnet/decode.sh --nj 20 --cmd run.pl --mem 4G --nnet exp/dnn4_pretrain-dbn_dnn_smbr/1.nnet --acwt 0.2 exp/tri3/graph data-fmllr-tri3/test exp/dnn4_pretrain-dbn_dnn_smbr/decode_test_it1

steps/nnet/decode.sh --nj 20 --cmd run.pl --mem 4G --nnet exp/dnn4_pretrain-dbn_dnn_smbr/1.nnet --acwt 0.2 exp/tri3/graph data-fmllr-tri3/dev exp/dnn4_pretrain-dbn_dnn_smbr/decode_dev_it1

steps/nnet/decode.sh --nj 20 --cmd run.pl --mem 4G --nnet exp/dnn4_pretrain-dbn_dnn_smbr/6.nnet --acwt 0.2 exp/tri3/graph data-fmllr-tri3/test exp/dnn4_pretrain-dbn_dnn_smbr/decode_test_it6

steps/nnet/decode.sh --nj 20 --cmd run.pl --mem 4G --nnet exp/dnn4_pretrain-dbn_dnn_smbr/6.nnet --acwt 0.2 exp/tri3/graph data-fmllr-tri3/dev exp/dnn4_pretrain-dbn_dnn_smbr/decode_dev_it6

Success

============================================================================

                    Getting Results [see RESULTS file]

============================================================================

%WER 31.9 | 400 15057 | 72.1 19.9 8.0 4.0 31.9 100.0 | -0.802 | exp/mono/decode_dev/score_4/ctm_39phn.filt.sys

%WER 24.6 | 400 15057 | 79.4 15.5 5.0 4.1 24.6 99.5 | -0.142 | exp/tri1/decode_dev/score_10/ctm_39phn.filt.sys

%WER 23.0 | 400 15057 | 80.6 14.4 5.0 3.6 23.0 99.5 | -0.238 | exp/tri2/decode_dev/score_10/ctm_39phn.filt.sys

%WER 20.0 | 400 15057 | 83.0 12.5 4.5 3.0 20.0 99.3 | -0.580 | exp/tri3/decode_dev/score_10/ctm_39phn.filt.sys

%WER 23.3 | 400 15057 | 80.0 14.8 5.2 3.3 23.3 99.8 | -0.169 | exp/tri3/decode_dev.si/score_10/ctm_39phn.filt.sys

%WER 20.8 | 400 15057 | 82.1 12.6 5.3 2.9 20.8 99.8 | -0.519 | exp/tri4_nnet/decode_dev/score_5/ctm_39phn.filt.sys

%WER 18.0 | 400 15057 | 84.4 11.1 4.5 2.4 18.0 99.3 | -0.150 | exp/sgmm2_4/decode_dev/score_10/ctm_39phn.filt.sys

%WER 18.1 | 400 15057 | 85.4 11.2 3.4 3.5 18.1 98.5 | -0.389 | exp/sgmm2_4_mmi_b0.1/decode_dev_it1/score_7/ctm_39phn.filt.sys

%WER 18.2 | 400 15057 | 84.5 11.2 4.3 2.7 18.2 99.0 | -0.155 | exp/sgmm2_4_mmi_b0.1/decode_dev_it2/score_10/ctm_39phn.filt.sys

%WER 18.2 | 400 15057 | 84.9 11.2 3.9 3.1 18.2 99.0 | -0.231 | exp/sgmm2_4_mmi_b0.1/decode_dev_it3/score_9/ctm_39phn.filt.sys

%WER 18.3 | 400 15057 | 84.5 11.3 4.2 2.8 18.3 99.0 | -0.170 | exp/sgmm2_4_mmi_b0.1/decode_dev_it4/score_10/ctm_39phn.filt.sys

%WER 17.2 | 400 15057 | 85.4 10.5 4.1 2.6 17.2 99.5 | -1.107 | exp/dnn4_pretrain-dbn_dnn/decode_dev/score_4/ctm_39phn.filt.sys

%WER 17.2 | 400 15057 | 85.6 10.5 3.9 2.8 17.2 99.3 | -1.070 | exp/dnn4_pretrain-dbn_dnn_smbr/decode_dev_it1/score_4/ctm_39phn.filt.sys

%WER 17.1 | 400 15057 | 86.1 10.6 3.3 3.3 17.1 99.3 | -1.095 | exp/dnn4_pretrain-dbn_dnn_smbr/decode_dev_it6/score_4/ctm_39phn.filt.sys

%WER 16.7 | 400 15057 | 86.1 10.9 3.0 2.8 16.7 98.8 | -0.124 | exp/combine_2/decode_dev_it1/score_6/ctm_39phn.filt.sys

%WER 16.8 | 400 15057 | 85.8 11.0 3.3 2.6 16.8 99.0 | -0.025 | exp/combine_2/decode_dev_it2/score_7/ctm_39phn.filt.sys

%WER 16.9 | 400 15057 | 86.0 11.0 3.0 2.8 16.9 98.8 | -0.121 | exp/combine_2/decode_dev_it3/score_6/ctm_39phn.filt.sys

%WER 16.8 | 400 15057 | 85.8 11.0 3.2 2.7 16.8 98.8 | -0.035 | exp/combine_2/decode_dev_it4/score_7/ctm_39phn.filt.sys

%WER 32.0 | 192 7215 | 71.3 19.3 9.5 3.3 32.0 100.0 | -0.425 | exp/mono/decode_test/score_5/ctm_39phn.filt.sys

%WER 25.7 | 192 7215 | 77.9 16.8 5.4 3.5 25.7 100.0 | -0.105 | exp/tri1/decode_test/score_10/ctm_39phn.filt.sys

%WER 24.3 | 192 7215 | 79.4 15.1 5.5 3.7 24.3 100.0 | -0.252 | exp/tri2/decode_test/score_10/ctm_39phn.filt.sys

%WER 21.5 | 192 7215 | 81.6 13.5 4.9 3.1 21.5 99.5 | -0.583 | exp/tri3/decode_test/score_10/ctm_39phn.filt.sys

%WER 24.1 | 192 7215 | 79.0 15.5 5.5 3.1 24.1 99.5 | -0.193 | exp/tri3/decode_test.si/score_10/ctm_39phn.filt.sys

%WER 23.0 | 192 7215 | 80.1 13.7 6.2 3.0 23.0 99.5 | -0.481 | exp/tri4_nnet/decode_test/score_5/ctm_39phn.filt.sys

%WER 19.4 | 192 7215 | 83.5 12.2 4.4 2.9 19.4 100.0 | -0.326 | exp/sgmm2_4/decode_test/score_8/ctm_39phn.filt.sys

%WER 19.6 | 192 7215 | 84.1 12.1 3.8 3.7 19.6 100.0 | -0.485 | exp/sgmm2_4_mmi_b0.1/decode_test_it1/score_7/ctm_39phn.filt.sys

%WER 19.9 | 192 7215 | 83.5 12.4 4.1 3.4 19.9 99.5 | -0.345 | exp/sgmm2_4_mmi_b0.1/decode_test_it2/score_8/ctm_39phn.filt.sys

%WER 20.0 | 192 7215 | 82.9 12.5 4.7 2.9 20.0 99.5 | -0.198 | exp/sgmm2_4_mmi_b0.1/decode_test_it3/score_10/ctm_39phn.filt.sys

%WER 20.1 | 192 7215 | 83.5 12.5 4.0 3.5 20.1 99.5 | -0.365 | exp/sgmm2_4_mmi_b0.1/decode_test_it4/score_8/ctm_39phn.filt.sys

%WER 18.4 | 192 7215 | 83.8 11.2 5.0 2.2 18.4 99.5 | -0.607 | exp/dnn4_pretrain-dbn_dnn/decode_test/score_6/ctm_39phn.filt.sys

%WER 18.4 | 192 7215 | 84.6 11.2 4.2 3.0 18.4 99.5 | -1.156 | exp/dnn4_pretrain-dbn_dnn_smbr/decode_test_it1/score_4/ctm_39phn.filt.sys

%WER 18.4 | 192 7215 | 84.8 11.3 3.9 3.2 18.4 99.0 | -0.787 | exp/dnn4_pretrain-dbn_dnn_smbr/decode_test_it6/score_5/ctm_39phn.filt.sys

%WER 18.4 | 192 7215 | 84.5 12.0 3.5 2.9 18.4 99.5 | -0.103 | exp/combine_2/decode_test_it1/score_6/ctm_39phn.filt.sys

%WER 18.3 | 192 7215 | 84.7 11.8 3.5 3.0 18.3 99.5 | -0.116 | exp/combine_2/decode_test_it2/score_6/ctm_39phn.filt.sys

%WER 18.3 | 192 7215 | 84.8 11.8 3.4 3.1 18.3 99.5 | -0.121 | exp/combine_2/decode_test_it3/score_6/ctm_39phn.filt.sys

%WER 18.2 | 192 7215 | 85.0 11.8 3.2 3.3 18.2 99.5 | -0.287 | exp/combine_2/decode_test_it4/score_5/ctm_39phn.filt.sys

============================================================================

Finished successfully on Fri Sep 15 17:39:16 CST 2017

============================================================================

kaldi的TIMIT实例三的更多相关文章

  1. kaldi的TIMIT实例一

    TIMIT语音库是IT和MIT合作音素级别标注的语音库,用于自动语音识别系统的发展和评估,包括来自美式英语,8个地区方言,630个人. 每个人读10个句子,每个发音都是音素级别.词级别文本标注,16k ...

  2. kaldi的TIMIT实例二

    ============================================================================ MonoPhone Training & ...

  3. C语言库函数大全及应用实例三

    原文:C语言库函数大全及应用实例三 [编程资料]C语言库函数大全及应用实例三 函数名: ecvt 功 能: 把一个浮点数转换为字符串 用 法: char ecvt(double value, int ...

  4. 登陆验证系统实例-三种（cookie,session,auth）

    登陆验证 因为http协议是无状态协议,但是我们有时候需要这个状态,这个状态就是标识 前端提交from表单,后端获取对应输入值,与数据库对比,由此对象设置一个标识,该对象 在别的视图的时候,有此标识, ...

  5. ZYNQ入门实例——三种GPIO应用、中断系统及软硬件交叉触发调试

    一.前言 Xlinx的ZYNQ系列SOC集成了APU.各种专用外设资源和传统的FPGA逻辑,为ARM+FPGA的应用提供助力,降低功耗和硬件设计难度的同时极大提高两者间传输的带宽.之前在研究生课题中使 ...

  6. PCIE_DMA实例三：Xilinx 7系列(KC705/VC709)FPGA的EDK仿真

    一:前言 好久没写博客了,前段时间有个朋友加微信请教关于PCIe的一些学习方法.本人也不是专家,只是略知一些皮毛.对于大家反馈的问题未必能一一解答,但一定知无不言.以后我会常来博客园看看,大家可以把问 ...

  7. IntelliJ IDEA 2017版 spring-boot 2.0.5 邮件发送简单实例 (三)

    一.搭建SpringBoot项目 详见此文:https://www.cnblogs.com/liuyangfirst/p/8298588.html 注意: 需要添加mail依赖的包,同时还添加了lom ...

  8. shell脚本实例三

    练习一:获得连通主机的ip和hostname1.脚本编写 vim  checkhost.sh #!/bin/bashAuto_conn(){/usr/bin/expect << EOFse ...

  9. Java 实例 - 三目条件运算符

    package guyu.day0903; /** * @Author: Fred * @Date: 2020/9/3 10:59 */ public class Demo01 { public st ...

随机推荐

  1. JS判断图片是否加载完成 背景图404 快到碗里来

    面对这个问题 我最多做到表面笑嘻嘻 …… 真不知道测试怎么那么…… 啥都能给你测出来 有的没的都能给你测出来 算了算了  谁让本仙女本精灵本可爱温柔大方善解人意呢 …呵呵呵 ————————————正 ...

  2. js二分算法排序

    var arr = [1,2,3,5,10,15,23,35,67,76,78,89,100];var len = arr.length;for (var i = 1; i < len; i++ ...

  3. 为nginx创建windows服务自启动

    1.下载最新版的 Windows Service Wrapper 程序,比如我下载的名称是 "winsw-1.9-bin.exe",然后,把它命名成你想要的名字(比如: " ...

  4. GUI学习之八——复选框QCheckBox的学习总结

    一.描述 a.QCheckBox一般用于给用户提供若干选项中多个选择时的使用 b.控件左侧有一个方框来显示控件被选中. c.复选框是有三种状态的 二.使用 1.创建 复选框的创建和常规的按钮创建方式是 ...

  5. -bash: /etc/profile: line 11: syntax error near unexpected token `$'{\r''报错问题解决

    在Linux系统配置Java环境变量之后执行 source /etc/profile指令报:-bash: /etc/profile: line 11: syntax error near unexpe ...

  6. 文件操作命令(TYPE)

    type 命令: // 描述: 显示文本文件的内容. 但不能进行修改. // 语法: type [<Drive>:][<Path>]<FileName> // 参数 ...

  7. 导入别人的项目eclipse 出现乱码 该如何处理

  8. 关于TypeError: strptime() argument 1 must be str, not bytes解析

    关于TypeError: strptime() argument 1 must be str, not bytes解析   在使用datetime.strptime(s,fmt)来输出结果日期结果时, ...

  9. LOJ-10096（强连通+bfs）

    题目链接:传送门 思路: 强连通缩点,重建图,然后广搜找最长路径. #include<iostream> #include<cstdio> #include<cstrin ...

  10. #pragma常用预处理指令

    #pragma pack(1):1字节对齐#pragma once:指定头文件被编译一次#pragma message("message"):编译时输出message文本#prag ...