Vladislavs Dovgalecs
Vladislavs Dovgalecs
@TompsonL In principle yes, in practice inference in such graphs would be *ahem* difficult. Even this triangular chain CRF is fairly expensive (the way it is implemented). Unfortunately I am...
@TompsonL I would try out a neural architecture - sequence model as for NER for "h" and classification on top of that for "y".
You might be interested in https://github.com/Oneplus/partial-crfsuite
You can do something like this: cat train.txt | crfsuite learn -m model -
The thing is that during training much more memory is quested than just fitting your dataset in the memory. For this big datasets, I suggest to use online algorithms. I...
The data format is similar to that of CRFSuite, except spaces are used to separate features. VW also introduces feature spaces. The following is a training example for sequence tagging...
The colon character is a special one. The value after colon is expected to be a scale value (float).
Can you define explicitly the labels and the features?
@dexception Did you read the documentation on the format? Did you check out the simple test file and did you try to run it?
@dexception Yes, this is the right example. Do you understand how it is built?