1 2 3 4 5 6 7
make svm faster (cuda)? other learning frameworks target side rule ngram feature template decoder meta-parameters test sa-extract -> leave-one-out? rerank while sgd?