def test(model_props=None,
model_name=None,
weights_file='best_weights',
dataset_name='test',
save_output=True,
save_scores=False):
if model_props is None:
model_props = model_properties.MentionRankingProps(
name=model_name,
load_weights_from=model_name,
weights_file=weights_file)
print "Loading data"
vectors = np.load(directories.RELEVANT_VECTORS + 'word_vectors.npy')
dataset = datasets.DocumentBatchedDataset(dataset_name,
model_props,
with_ids=True)
docs = utils.load_pickle(directories.DOCUMENTS + dataset_name +
'_docs.pkl')
stats = {}
print "Building model"
model, _ = pairwise_models.get_model(dataset, vectors, model_props)
print "Evaluating model on", dataset_name
evaluate_model(dataset,
docs,
model,
model_props,
stats,
save_output=save_output,
save_scores=save_scores)
timer.clear()
utils.write_pickle(stats, model_props.path + dataset_name + "_results.pkl")
def main(model_props=None, cluster_props=None):
if model_props is None:
model_props = model_properties.MentionRankingProps()
if cluster_props is None:
cluster_props = model_properties.ClusterRankingProps()
directories.set_model_name('classification')
model_props.load_weights_from = None
model_props.set_mode('classification')
pairwise_learning.main(model_props=model_props, n_epochs=150)
directories.set_model_name('top_pairs')
model_props.set_mode('top_pairs')
model_props.load_weights_from = 'classification'
model_props.weights_file = 'weights_140'
timer.clear()
pairwise_learning.main(model_props=model_props, n_epochs=50)
directories.set_model_name('ranking')
model_props.set_mode('ranking')
model_props.load_weights_from = 'top_pairs'
model_props.weights_file = 'weights_40'
timer.clear()
pairwise_learning.main(model_props=model_props, n_epochs=50)
model_props.load_weights_from = 'ranking'
pairwise_learning.main(model_props=model_props, write_scores=True, test_only=True)
pairwise_learning.main(model_props=model_props, write_scores=True, test_only=True,
validate="test")
clustering_preprocessing.main('ranking')
cluster_props.load_weights_from = 'ranking'
cluster_props.weights_file = 'weights_40'
timer.celar()
clustering_learning.main(cluster_props)
def load_docs(dataset_name, word_vectors):
return (dataset.Dataset(dataset_name,
model_properties.MentionRankingProps(),
word_vectors),
zip(
util.load_pickle(directories.DOCUMENTS + dataset_name +
'_docs.pkl'),
util.load_pickle(directories.ACTION_SPACE + dataset_name +
'_action_space.pkl')))
def train_best_model():
train_and_test_pairwise(model_properties.MentionRankingProps(),
mode='reward_rescaling')
def emnlp2016():
model_props = model_properties.MentionRankingProps()
train_and_test_pairwise(model_props, mode='ranking')
train_and_test_pairwise(model_props, mode='reinforce')
train_and_test_pairwise(model_props, mode='reward_rescaling')
def acl2016():
model_props = model_properties.MentionRankingProps()
train_pairwise(model_props)
make_predictions(model_props, ["train", "dev", "test"], save_scores=True)
train_clustering(model_properties.ClusterRankingProps())