There is really some stuff that I don’t understand.
Still can’t run more than two intents, waiting for a new docker version, but I’m really afraid about running lot of intents and slors
No intent files, just in sentences tab I enter:
[GetTime]
quelle heure est-il
il est quelle heure
[EntityStateValueJeedom]
c'est quelle (musique | chanson){device_name:musique} qu'on écoute ?
quelle est cette (musique | chanson){device_name:musique} ?
c'est quelle (musique | chanson){device_name:musique} çà ?
c'est quoi cette (musique | chanson){device_name:musique} ?
Training failed: <Task: language_model:intent_model>: TaskFailed Command failed: 'ngrammake /profiles/fr/intent.fst.counts /profiles/fr/intent.fst.model' returned 1
But this works:
[GetTime]
quelle heure est-il
il est quelle heure
[EntityStateValueJeedom]
c'est quelle (musique | chanson){device_name:musique} qu'on écoute ?
quelle est cette (musique | chanson){device_name:musique} ?
Or this:
[EntityStateValueJeedom]
c'est quelle (musique | chanson){device_name:musique} qu'on écoute ?
quelle est cette (musique | chanson){device_name:musique} ?
c'est quelle (musique | chanson){device_name:musique} çà ?
c'est quoi cette (musique | chanson){device_name:musique} ?
I’ve done test like with set of complete sentences or setting just one or two to produce same result. Don’t make a difference, always both forms are ok or nok.
I’m seeing this error too with this also coming out of opengrm:
FATAL: NGramKatz: No discount bin for discounting
I’ll be digging into this issue more. I can’t even guess what’s wrong yet, except that it has something to do with the generation of the language model.
When training, I get the following “Command failed: ‘ngramcount /home/pi/.config/rhasspy/profiles/de/intent.fst /home/pi/.config/rhasspy/profiles/de/intent.fst.counts’ returned 1” error:
INFO:main:Starting training
DEBUG:DialogueManager:asleep -> training_sentences
[2020-01-22 15:44:32,645] 127.0.0.1:57076 GET /api/unknown-words 1.1 200 2 50321
INFO:quart.serving:127.0.0.1:57076 GET /api/unknown-words 1.1 200 2 50321
DEBUG:train:Parsing ini file(s): [’/home/pi/.config/rhasspy/profiles/de/sentences.ini’]
DEBUG:rhasspynlu.ini_jsgf:Loaded ini file
– ini_graph
– intent_fst
. language_model:intent_counts
ngramcount: /lib/arm-linux-gnueabihf/libm.so.6: version GLIBC_2.27' not found (required by ngramcount) ngramcount: /lib/arm-linux-gnueabihf/libm.so.6: version GLIBC_2.27’ not found (required by /home/pi/rhasspy/.venv/lib/libngram.so.134)
TaskFailed - taskid:language_model:intent_counts
Command failed: ‘ngramcount /home/pi/.config/rhasspy/profiles/de/intent.fst /home/pi/.config/rhasspy/profiles/de/intent.fst.counts’ returned 1
DEBUG:DialogueManager:training_sentences -> ready
INFO:DialogueManager:Automatically listening for wake word
DEBUG:DialogueManager:ready -> asleep
DEBUG:InboxActor: -> stopped
ERROR:main:Training failed: <Task: language_model:intent_counts>: TaskFailed
Command failed: ‘ngramcount /home/pi/.config/rhasspy/profiles/de/intent.fst /home/pi/.config/rhasspy/profiles/de/intent.fst.counts’ returned 1
Traceback (most recent call last):
File “/home/pi/rhasspy/.venv/lib/python3.7/site-packages/quart/app.py”, line 1471, in full_dispatch_request
result = await self.dispatch_request(request_context)
File “/home/pi/rhasspy/.venv/lib/python3.7/site-packages/quart/app.py”, line 1519, in dispatch_request
return await handler(**request_.view_args)
File “app.py”, line 610, in api_train
raise Exception(f"Training failed: {result.reason}")
Exception: Training failed: <Task: language_model:intent_counts>: TaskFailed Command failed: ‘ngramcount /home/pi/.config/rhasspy/profiles/de/intent.fst /home/pi/.config/rhasspy/profiles/de/intent.fst.counts’ returned 1
[2020-01-22 15:44:32,855] 127.0.0.1:57072 POST /api/train 1.1 500 209 288358
INFO:quart.serving:127.0.0.1:57072 POST /api/train 1.1 500 209 288358
DEBUG:InboxActor: -> stopped
[2020-01-22 15:44:33,042] 127.0.0.1:57072 GET /api/problems 1.1 200 292 35427
INFO:quart.serving:127.0.0.1:57072 GET /api/problems 1.1 200 292 35427
I do not know if there is a connection but I changed from a docker install to venv recently. intent.fst, intent.fst.model and intent.fst.count exist and my sentence.ini looks like this