Many thanks, Kenneth. Here it is.
....... Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0x7ffff6ae1700 (LWP 11397)] Moses::ChartManager::CalcNBest (this=<optimised out>, count=100, ret=..., onlyDistinct=false) at moses/src/ChartManager.cpp:217 217 const HypoList topHypos = *(lastCell.GetSortedHypotheses(*w)); (gdb) backtrace #0 Moses::ChartManager::CalcNBest (this=<optimised out>, count=100, ret=..., onlyDistinct=false) at moses/src/ChartManager.cpp:217 #1 0x000000000041027b in TranslationTask::Run (this=0x2a67f50) at moses-chart-cmd/src/Main.cpp:122 #2 0x000000000052e5b1 in Moses::ThreadPool::Execute (this=0x7fffffffd960) at moses/src/ThreadPool.cpp:58 #3 0x000000000068d034 in thread_proxy () #4 0x00007ffff71aae9a in start_thread (arg=0x7ffff6ae1700) at pthread_create.c:308 #5 0x00007ffff6ed7cbd in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:112 #6 0x0000000000000000 in ?? () It seems something is checking at address 0. Hope the info helps you (and me). Regards, Guchun On 25 March 2013 10:47, Guchun Zhang <[email protected]> wrote: > Hi, > > Hope someone can give me some clues for this issue. Basically, tuning dies > in the first run and the error is > > */var/spool/gridengine/execd/ubuntu11/job_scripts/82585: line 9: 23008 > Segmentation fault (core dumped) > /home/guchun/Work/moses/mosesdecoder/bin/moses_chart -w -0.285714 -lm > 0.142857 -tm 0.057143 0.057143 0.057143 0.057143 0.057143 0.285714 -config > /home/guchun/Work/moses/tests/exp-german/model/DE_DE-EN_US/tuning/moses.filtered.ini.1 > -inputtype 0 -n-best-list > /home/guchun/Work/moses/tests/exp-german/model/DE_DE-EN_US/tuning/tmp.1/tmp27625/run1.best100.out.split27625-aa > 100 -input-file input.tc.1.split27625-aa > > /home/guchun/Work/moses/tests/exp-german/model/DE_DE-EN_US/tuning/tmp.1/tmp27625/input.tc.1.split27625-aa.trans > * > > The language pair is DE-EN. I am using Berkeley Parser for both languages. > The Moses version is about 6 months old. If I omit *n-best-list* and * > inputtype*, the decoder runs OK. But since *n-best-list* is essential to > tuning, what can I do to solve this issue? > > Many thanks, > Guchun >
_______________________________________________ Moses-support mailing list [email protected] http://mailman.mit.edu/mailman/listinfo/moses-support
