I was trying to generate trie for new binary file i have generated from added vocabulary.
When I am trying to compile its giving error.
/native_client# g++ generate_trie.cpp
In file included from generate_trie.cpp:5:0:
ctcdecode/scorer.h:9:10: fatal error: lm/enumerate_vocab.hh: No such file or directory
#include "lm/enumerate_vocab.hh"
^~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
Please help, thanks in advance!
lissyx
((slow to reply) [NOT PROVIDING SUPPORT])
2
Please, read the documentation, you are not building it correctly.
1 Like
lissyx
((slow to reply) [NOT PROVIDING SUPPORT])
3
generate_trie is distributed as part of native_client.tar.xz, you should not need to rebuild it.
Hi @lissyx , I have generated my binaries with added vocabularies. and also generated the trie. but during training with the new trie and new binaries its showing segmentation fault (core dump) ?
lissyx
((slow to reply) [NOT PROVIDING SUPPORT])
5
Please share more logs on the crash, ensure you are running ds_ctcdecoder version matching deepspeech training code and python version, and cross-check with released lm and trie file to verify it is not yours.
Use standard file APIs to delete files with this prefix.
Epoch 0 | Training | Elapsed Time: 3:06:01 | Steps: 569 | Loss: 27.668679
Epoch 0 | Validation | Elapsed Time: 0:01:12 | Steps: 32 | Loss: 57.548914 | Dataset: /home/rbeigcn1134841d/Downloads/indian_dev.csv
I Saved new best validating model with loss 57.548914 to: /home/rbeigcn1134841d/Desktop/mark1/DeepSpeech/deepspeech-0.6.1-checkpoint/best_dev-236060
I FINISHED optimization in 3:07:15.253444
INFO:tensorflow:Restoring parameters from /home/rbeigcn1134841d/Desktop/mark1/DeepSpeech/deepspeech-0.6.1-checkpoint/best_dev-236060
I0204 16:53:11.407811 140016252487488 saver.py:1284] Restoring parameters from /home/rbeigcn1134841d/Desktop/mark1/DeepSpeech/deepspeech-0.6.1-checkpoint/best_dev-236060
I Restored variables from best validation checkpoint at /home/rbeigcn1134841d/Desktop/mark1/DeepSpeech/deepspeech-0.6.1-checkpoint/best_dev-236060, step 236060
Testing model on /home/rbeigcn1134841d/Downloads/indian_test.csv
Test epoch | Steps: 0 | Elapsed Time: 0:00:00 Fatal Python error: Fatal Python error: Segmentation faultSegmentation fault
Thread 0x
00007f5735fff700 (most recent call first):
File Segmentation fault (core dumped)
lissyx
((slow to reply) [NOT PROVIDING SUPPORT])
7
Right, so it’s test step, it’s likely to be mismatched ds_ctcdecoder. Please reinstall / upgrade it: pip install --upgrade $(python util/taskcluster.py --decoder).
Also ,
If i use lm.binary in default package and my trie its giving core-dump. But if I use my binaries and trie given in the default package its working. Not sure why?
Yes default lm.binary and trie are working perfectly fine
Ok will check the generate_lm script and see the docs.
lissyx
((slow to reply) [NOT PROVIDING SUPPORT])
20
Weird. If you are on v0.6.1, you should not have that tag. This shows you are on master, so you’re going to have troubles if you don’t stick to matching versions.