Error: Trie file version mismatch (4 instead of expected 5). Update your trie file for deepspeech 0.6.0a15

Deep Speech
#1

i am doing training my deepspeech model after building lm.binary and trie using kenlm.
i am using deepspeech==0.6.0a15
ds-ctcdecoder==0.6.0a15
python 3.7.3

i am getting error Error: Trie file version mismatch (4 instead of expected 5). Update your trie file.
terminate called after throwing an instance of ‘int’
Fatal Python error: Aborted.

#2

Have you built it using the proper versions ? The error would suggest that no.

#3

Hi @lissyx

i have built trie and lm.binary using the following after installing kenlm:

./lmplz --text vocabulary.txt --arpa words.arpa --o 5 --discount_fallback --temp_prefix /tmp/

./build_binary -T -s words.arpa lm.binary

./generate_trie alphabet.txt lm.binary trie

Is it not the proper way, pls advise on built it using the proper versions.

#4

For 0.6.0a15 the instructions on building a language model are here[1] and differ from what you have done. In particular you do not use the -a and -q flags. In addition, make sure you are using the proper 0.6.0a15 version of generate_trie

#5

i solved the problem by downloading generate_trie
python util/taskcluster.py --target . --branch v0.6.0.

thanks for your help.