faster-rnnlm
faster-rnnlm copied to clipboard
yandex
the binary format of output
Hi, Everyone. I want to ask you some questions. How to convert the binary format of output into the text format of output? Are there correspond parameter settings? Thank you.
Hi! There is not flag to use text format. All the information about the network is stored in NNet class. Once it is constructed, you can dump it in a way you like. For instance, you can dump main_nnet instance here: https://github.com/yandex/faster-rnnlm/blob/master/faster-rnnlm/rnnlm.cc#L759
Merry Christmas! I want to use the kaldi tools ,but I don't konw how to use the binary format in decoding.Can you tell me how to use it if possible. Thank you!
In kaldi faster-rnnln is used for rescoring only. For instance, checkout librispeech recipe: https://github.com/kaldi-asr/kaldi/blob/master/egs/librispeech/s5/run.sh#L333
On Fri, Dec 25, 2015 at 4:10 AM, xiaofeixia1234 [email protected] wrote:
Merry Christmas! I want to use the kaldi tools ,but I don't konw how to use the binary format in decoding.Can you tell me how to use it if possible. Thank you!
— Reply to this email directly or view it on GitHub https://github.com/yandex/faster-rnnlm/issues/11#issuecomment-167174150.
With best wishes, Anton Bakhtin
Hi ,I also have some problems . /home/kaldi-trunk/tools/faster-rnnlm-master/faster-rnnlm/rnnlm WARNING: unknown option 'debug' Read the vocabulary: 16085 words Restoring existing nnet Constructing RNN: layer_size=400, layer_type=sigmoid, layer_count=1, maxent_hash_size=999988365, maxent_order=4, vocab_size=16085, use_nce=0
My option is time $taskset_cmd $1 -rnnlm $basedir/models/rnnlm -train $data/train -valid $data/valid -hidden $hidden_size -use-cuda 1 -bptt 4 -bptt-skip 20 -direct 1000 -direct-order 4 -alpha 0.01 -threads $threads ${3:-} $1 -rnnlm $basedir/models/rnnlm -test $data/test -nce-accurate-test 1 2>&1 > /dev/null | grep "Test entropy" | cat
I don't know which option is error! Thank you!