The error is happening when I use my ReloadingLM class, which inherits from LanguageModelKen. The problem only appears when I start with one model and then reload a different model.
When the model is reloaded, the code calls LanguageModelKen::LoadModel again, resetting m_ngram. On Tue, Mar 22, 2016 at 1:30 PM, Lane Schwartz <[email protected]> wrote: > Here's the backtrace. It's weird. I get this error under certain > conditions, but not under others. > > Files are attached. > > > > On Tue, Mar 22, 2016 at 4:20 AM, Kenneth Heafield <[email protected]> > wrote: > >> Hi, >> >> I don't see a problem with this in principle. The error means >> that the >> calling code provided an out-of-range word id. Can I have a backtrace >> after compiling with debug? >> >> Kenneth >> >> On 03/21/2016 11:34 PM, Lane Schwartz wrote: >> > Hi, >> > >> > I have a tiny LM that is giving me some problems. The LM wasn't trained >> > with any of the usual toolkits, so it may violate some assumptions that >> > the KenLM code in Moses makes. >> > >> > I'm able to successfully query the LM using bin/query, but when I use >> > the LM within Moses I get the following error: >> > >> > Reading dd_lm0.arpa >> > >> >> ----5---10---15---20---25---30---35---40---45---50---55---60---65---70---75---80---85---90---95--100 >> > >> >> **************************************************************************************************** >> > mosesserver: lm/model.cc:250: lm::FullScoreReturn >> > lm::ngram::detail::GenericModel<Search, >> > VocabularyT>::ScoreExceptBackoff(const WordIndex*, const WordIndex*, >> > lm::WordIndex, lm::ngram::State&) const [with Search = >> > lm::ngram::detail::HashedSearch<lm::ngram::BackoffValue>; >> > VocabularyT = lm::ngram::ProbingVocabulary; lm::WordIndex = unsigned >> > int]: Assertion `new_word < vocab_.Bound()' failed. >> > >> > >> > And here's the LM: >> > >> > \data\ >> > ngram 1=22 >> > ngram 2=21 >> > >> > >> > \1-grams: >> > 0.-99 >> > 0:-99 >> > 0</s>-99 >> > -99<s>-99 >> > -99<unk> >> > -4.30102999566a-99 >> > -5.00005affect >> > 0certain-99 >> > -4.30102999566have-99 >> > -4.30102999566impact-99 >> > 0is-99 >> > -4.30102999566negative-99 >> > -5.00005negatively >> > 0new-99 >> > -4.30102999566on-99 >> > 0one-99 >> > 0provisions-99 >> > 0the-99 >> > 0these-99 >> > 0thing-99 >> > 0turnout-99 >> > 0will-99 >> > >> > >> > \2-grams: >> > 0. </s> >> > 0: these >> > 0<s> one >> > -4.30102999566a negative >> > -5.00005affect negatively >> > 0certain : >> > -4.30102999566have a >> > -4.30102999566impact on >> > 0is certain >> > -4.30102999566negative impact >> > -5.00005negatively the >> > 0new provisions >> > -4.30102999566on the >> > 0one thing >> > 0provisions will >> > 0the turnout >> > 0these new >> > 0thing is >> > 0turnout . >> > -5.00005will affect >> > -4.30102999566will have >> > \end\ >> > >> > >> > >> > _______________________________________________ >> > Moses-support mailing list >> > [email protected] >> > http://mailman.mit.edu/mailman/listinfo/moses-support >> > >> _______________________________________________ >> Moses-support mailing list >> [email protected] >> http://mailman.mit.edu/mailman/listinfo/moses-support >> > > > > -- > When a place gets crowded enough to require ID's, social collapse is not > far away. It is time to go elsewhere. The best thing about space travel > is that it made it possible to go elsewhere. > -- R.A. Heinlein, "Time Enough For Love" > -- When a place gets crowded enough to require ID's, social collapse is not far away. It is time to go elsewhere. The best thing about space travel is that it made it possible to go elsewhere. -- R.A. Heinlein, "Time Enough For Love"
_______________________________________________ Moses-support mailing list [email protected] http://mailman.mit.edu/mailman/listinfo/moses-support
