def train_test(model=None,
train_examples=[],
test_examples=[],
metrics=standard_metrics(),
training_metric=SemanticsAccuracyMetric(),
seed=None,
print_examples=False):
# print_grammar(model.grammar)
# print
print('%d training examples, %d test examples' %
(len(train_examples), len(test_examples)))
# 'Before' test
model.weights = defaultdict(float) # no weights
evaluate_model(model=model,
examples=train_examples,
examples_label='train',
metrics=metrics,
print_examples=print_examples)
evaluate_model(model=model,
examples=test_examples,
examples_label='test',
metrics=metrics,
print_examples=print_examples)
# Train
model = latent_sgd(model,
train_examples,
training_metric=training_metric,
seed=seed)
# 'After' test
evaluate_model(model=model,
examples=train_examples,
examples_label='train',
metrics=metrics,
print_examples=print_examples)
evaluate_model(model=model,
examples=test_examples,
examples_label='test',
metrics=metrics,
print_examples=print_examples)
def interact(domain, example_input=None, T=10):
import readline
model = domain.model()
model = latent_sgd(model=model,
examples=domain.train_examples(),
training_metric=domain.training_metric(),
T=T)
print('\nHello! Enter a query%s:' % (', such as "%s"' % example_input if example_input else ''))
while True:
try:
input = input('>>> ')
except EOFError:
print('\nBye!')
return
example = Example(input=input)
parses = model.parse_input(input)
if parses:
print_parses(example, parses)
else:
print('No parse!')
def interact(domain, example_input=None, T=10):
import readline
model = domain.model()
model = latent_sgd(model=model,
examples=domain.train_examples(),
training_metric=domain.training_metric(),
T=T)
print('\nHello! Enter a query%s:' %
(', such as "%s"' % example_input if example_input else ''))
while True:
try:
query = input('>>> ')
except EOFError:
print('\nBye!')
return
example = Example(input=query)
parses = model.parse_input(query)
if parses:
print_parses(example, parses)
else:
print('No parse!')
def train_test(model=None,
train_examples=[],
test_examples=[],
metrics=standard_metrics(),
training_metric=SemanticsAccuracyMetric(),
seed=None,
print_examples=False):
# print_grammar(model.grammar)
# print
print('%d training examples, %d test examples' % (len(train_examples), len(test_examples)))
# 'Before' test
model.weights = defaultdict(float) # no weights
evaluate_model(model=model,
examples=train_examples,
examples_label='train',
metrics=metrics,
print_examples=print_examples)
evaluate_model(model=model,
examples=test_examples,
examples_label='test',
metrics=metrics,
print_examples=print_examples)
# Train
model = latent_sgd(model, train_examples, training_metric=training_metric, seed=seed)
# 'After' test
evaluate_model(model=model,
examples=train_examples,
examples_label='train',
metrics=metrics,
print_examples=print_examples)
evaluate_model(model=model,
examples=test_examples,
examples_label='test',
metrics=metrics,
print_examples=print_examples)
