Reranker Framework (ReFr)
Reranking framework for structure prediction and discriminative language modeling
|
Provides the reranker::PerceptronModel reranker class. More...
#include <vector>
#include <unordered_set>
#include <memory>
#include "candidate-set.H"
#include "dot-product.H"
#include "model.H"
#include "training-time.H"
#include "training-vector-set.H"
Go to the source code of this file.
Classes | |
class | reranker::PerceptronModel |
This class implements a perceptron model reranker. More... | |
class | reranker::PerceptronModel::DefaultUpdatePredicate |
The default update predicate for perceptron and perceptron-style models, which indicates to do a model update whenever the top-scoring candidate hypothesis under the current model differs from the oracle or “gold” candidate hypothesis. More... | |
class | reranker::PerceptronModel::DefaultUpdater |
The default update function for perceptron models. More... | |
Namespaces | |
reranker | |
Provides reranking models for discriminative modeling, with some special handling for discriminative language models. | |
Macros | |
#define | DEFAULT_MAX_EPOCHS_IN_DECLINE 5 |
Provides the reranker::PerceptronModel reranker class.
Definition in file perceptron-model.H.
#define DEFAULT_MAX_EPOCHS_IN_DECLINE 5 |
Definition at line 49 of file perceptron-model.H.