I used CRF++ to learn a model with around 6,000 sentences. The trained model has been tested and it proved to work perfectly. Then, I increased the number of sentences to 120,000 sentences. The problem is when I trained the model using the following command line
crf_learn template traindata model
After crf_learn.exe reading the training data, it suddenly stops working and an error message appears which closed the program and prevented it from completing the steps. the error message says crf_learn.exe has stopped working correctly. Windows will close the program.
I tried to make the training faster by using multi-threading and cuting-off the threshold for the features as follow:
crf_learn -f 30 -c 1.5 template traindata model
but the problem still appears and I could not create the trained model! Is there any one have an experience with CRF++? or Is there any way to increase the memory for executable file (.exe) that have already built using C++ when running it from Command line?