def main(argv):
flags = get_flags(argv, __doc__, argparser)
print_result(
get_total_events(
flags.table_id,
flags.label,
flags.action
),
flags.action,
)
## Standardize data
train_mean, train_std = get_data_info(train_images)
# train_images = standardize_data(train_images, train_mean, train_std)
# validation_images = standardize_data(validation_images, train_mean, train_std)
# test_images = standardize_data(test_images, train_mean, train_std)
data = DataSet(train_images, train_labels)
test = DataSet(test_images, test_labels)
val = DataSet(validation_images, validation_labels)
return data, test, val
if __name__ == '__main__':
FLAGS = utils.get_flags()
## Set random seed for tensorflow and numpy operations
tf.set_random_seed(FLAGS.seed)
np.random.seed(FLAGS.seed)
data, test, _ = import_mnist()
## Here we define a custom loss for dgp to show
error_rate = losses.ZeroOneLoss(data.Dout)
## Likelihood
like = likelihoods.Softmax()
## Optimizer
optimizer = utils.get_optimizer(FLAGS.optimizer, FLAGS.learning_rate)
# Load CS objects for single stimuli plotting
iter_vars_dims = []
for iter_var in iter_vars:
iter_vars_dims.append(len(iter_vars[iter_var]))
print('Loading object list for single stimulus plot...'),
CS_object_array = load_aggregated_object_list(iter_vars_dims,
data_flag)
print('...loaded.')
# Plot signal and inset
ax['signal_%s' % Kk_split_idx] = \
signal_plot(ax['signal_%s' % Kk_split_idx])
ax['signal_insert_%s' % Kk_split_idx] = \
signal_plot(ax['signal_insert_%s' % Kk_split_idx])
if Weber_idx == 1:
mark_inset(ax['signal_%s' % Kk_split_idx],
ax['signal_insert_%s' % Kk_split_idx],
loc1=3,
loc2=4,
fc="none",
ec="0.5")
save_signal_discrimination_weber_law_fig(fig)
if __name__ == '__main__':
data_flags = get_flags()
plot_signal_discrimination_weber_law(
data_flags,
axes_to_plot=[0, 1],
projected_variable_components=dict(normal_eps_tuning_width=15))
def get_answer(question, story, sch_flag=False):
flag_500 = story['sid'].startswith(
'mc500') # mctrain500 missing the sch data, changes pattern
qflags = utils.get_flags(question)
sch_flag = 'Sch' in question['type']
whole = nltk.word_tokenize(question['text'])
if not any(qflags[key] for key in qflags):
for i in whole:
qflags = utils.get_flags(i)
if qflags['who']:
#sentence selection:
#resolve anaphora if necesary
#similarity overlap , fallback to word overlap
answer, i = who_baseline(question, story, sch_flag=sch_flag)
best_dep = wn_extract(question, story, i, sch_flag=sch_flag)
answer = (best_dep if best_dep else answer) #'next code'
#answer = 'next code'
elif qflags['what']:
# distinguish between verb and noun and quote return type
# select sentence with similarity overlap as a first choice
# failing onto word overlap of sch if possible
return_type = utils.return_type(question)
answer, i = what_baseline(question,
story,
return_type,
sch_flag=sch_flag)
best_dep = wn_extract(question, story, i, sch_flag=sch_flag)
answer = (best_dep if best_dep else answer)
#answer = 'next code'
elif qflags['when']:
kw_adds = [
'Saturday', 'Sunday', 'Monday', 'Tuesday', 'Wednesday', 'Thursday',
'Friday'
]
kw_adds += pp_filter
answer, i = when_baseline(
question,
story,
kw_adds,
sch_flag,
)
best_dep = wn_extract(question, story, i, sch_flag=sch_flag)
answer = (best_dep if best_dep else answer) #'next code'
#answer = 'next code'
elif qflags['why']:
#add why answer triggers to the question when looking for overlap
kw_adds = why_filter
answer, i = when_baseline(question, story, kw_adds, sch_flag)
best_dep = wn_extract(question, story, i, sch_flag=sch_flag)
answer = (best_dep if best_dep else answer) #'next code'
#answer = 'next code'
elif qflags['where']:
kw_adds = ['in', 'where', 'at', 'front', 'back', 'outside', 'inside']
answer, i = when_baseline(question, story, kw_adds, sch_flag=sch_flag)
best_dep = wn_extract(question, story, i, sch_flag=sch_flag)
answer = (best_dep if best_dep else answer)
elif qflags['which']:
#question reformation
kw_adds = []
answer, i = when_baseline(question, story, kw_adds, sch_flag)
best_dep = wn_extract(question, story, i)
answer = (best_dep if best_dep else answer)
#answer = 'next code'
elif qflags['did']:
#question reformation
#simple overlap , look for 'nt in answer, make a score threshold
#based on the number of key words
kw_adds = []
answer, i = when_baseline(question, story, kw_adds, sch_flag=sch_flag)
best_dep = wn_extract(question, story, i)
answer = (best_dep if best_dep else answer)
answer = 'yes' if "'nt" not in answer else 'no'
#answer = 'next code'
elif qflags['how']:
#resovle whether adj or verb gerund return type
kw_adds = []
answer, i = when_baseline(question, story, kw_adds, sch_flag=sch_flag)
best_dep = wn_extract(question, story, i, sch_flag=sch_flag)
answer = (best_dep if best_dep else answer)
else:
#dialogues questions
#seriously word overlap
#just get the sentence then question reformation
kw_adds = []
answer, i = when_baseline(question, story, kw_adds, sch_flag=sch_flag)
best_dep = wn_extract(question, story, i)
answer = (best_dep if best_dep else answer)
#answer = 'next code'
return answer
