def _conversion_decimal(context, is_train, is_extrapolation):
"""E.g., "How many grams are in 5kg?"."""
dimension = random.choice(DIMENSIONS)
while True:
base_value, base_unit, target_value, target_unit = (
_sample_conversion_decimal(dimension, is_extrapolation))
if train_test_split.is_train(base_value) == is_train:
break
templates = [
'Ada berapa {target_name} di {base_value} {base_name}?',
'Apa {base_value} {base_name} dalam {target_name}?',
'Ubah {base_value} {base_name} to {target_name}.',
]
if base_unit.symbol is not None:
templates += [
'Ada berapa {target_name} di {base_value}{base_symbol}?',
'Apa {base_value}{base_symbol} dalam {target_name}?',
'Ubah {base_value}{base_symbol} ke {target_name}.',
]
template = random.choice(templates)
base_name = base_unit.name # Indonesia no have plural stuff
target_name = target_unit.name # Indonesia no have plural stuff
question = example.question(context,
template,
base_name=base_name,
base_symbol=base_unit.symbol,
base_value=base_value,
target_name=target_name)
return example.Problem(question=question, answer=target_value)
def _conversion_decimal(context, is_train, is_extrapolation):
"""E.g., "How many grams are in 5kg?"."""
dimension = random.choice(DIMENSIONS)
while True:
base_value, base_unit, target_value, target_unit = (
_sample_conversion_decimal(dimension, is_extrapolation))
if train_test_split.is_train(base_value) == is_train:
break
templates = [
'How many {target_name} are there in {base_value} {base_name}?',
'What is {base_value} {base_name} in {target_name}?',
'Convert {base_value} {base_name} to {target_name}.',
]
if base_unit.symbol is not None:
templates += [
'How many {target_name} are there in {base_value}{base_symbol}?',
'What is {base_value}{base_symbol} in {target_name}?',
'Convert {base_value}{base_symbol} to {target_name}.',
]
template = random.choice(templates)
base_name = pluralize(base_unit.name)
target_name = pluralize(target_unit.name)
question = example.question(
context,
template,
base_name=base_name,
base_symbol=base_unit.symbol,
base_value=base_value,
target_name=target_name)
return example.Problem(question=question, answer=target_value)
def time(is_train):
"""Questions for calculating start, end, or time differences."""
context = composition.Context()
start_minutes = random.randint(1, 24 * 60 - 1)
while True:
duration_minutes = random.randint(1, 12 * 60 - 1)
if train_test_split.is_train(duration_minutes) == is_train:
break
end_minutes = start_minutes + duration_minutes
def format_12hr(minutes):
"""Format minutes from midnight in 12 hr format."""
hours = (minutes // 60) % 24
minutes %= 60
am_pm = 'AM' if hours < 12 else 'PM'
hours = (hours - 1) % 12 + 1
return '{}:{:02} {}'.format(hours, minutes, am_pm)
start = format_12hr(start_minutes)
end = format_12hr(end_minutes)
which_question = random.randint(0, 3)
if which_question == 0:
# Question: What is start = end - duration?
template = random.choice([
'Berapa {duration} menit sebelum {end}?',
])
return example.Problem(question=example.question(
context, template, duration=duration_minutes, end=end),
answer=start)
elif which_question == 1:
# Question: What is end = start + duration?
template = random.choice([
'Berapa {duration} menit setelah {start}?',
])
return example.Problem(question=example.question(
context, template, duration=duration_minutes, start=start),
answer=end)
else:
# Question: What is duration = end - start?
template = random.choice([
'Berapa menit antara {start} dan {end}?',
])
return example.Problem(question=example.question(context,
template,
start=start,
end=end),
answer=duration_minutes)
def _sample_letter_bag(is_train, min_total):
"""Samples a "container of letters" and returns info on it."""
while True:
num_distinct_letters = random.randint(1, _MAX_DISTINCT_LETTERS)
num_letters_total = random.randint(
max(num_distinct_letters, min_total),
min(_MAX_TOTAL_LETTERS, num_distinct_letters * _MAX_LETTER_REPEAT))
letter_counts = combinatorics.uniform_positive_integers_with_sum(
num_distinct_letters, num_letters_total)
# Test/train split.
if (is_train is None or train_test_split.is_train(
sorted(letter_counts)) == is_train):
break
letters_distinct = random.sample(_LETTERS, num_distinct_letters)
weights = {i: 1 for i in range(num_letters_total)}
letters_with_repetition = []
for letter, count in zip(letters_distinct, letter_counts):
letters_with_repetition += [letter] * count
random.shuffle(letters_with_repetition)
random_variable = probability.DiscreteRandomVariable(
{i: letter
for i, letter in enumerate(letters_with_repetition)})
if random.choice([False, True]):
bag_contents = ''.join(letters_with_repetition)
else:
letters_and_counts = [
'{}: {}'.format(letter, count)
for letter, count in zip(letters_distinct, letter_counts)
]
bag_contents = '{' + ', '.join(letters_and_counts) + '}'
return LetterBag(weights=weights,
random_variable=random_variable,
letters_distinct=letters_distinct,
bag_contents=bag_contents)
def _conversion_fraction(context, is_train):
"""E.g., "How many grams are in three quarters of a kg?"."""
dimension = random.choice(DIMENSIONS)
# Limit probability of giving zero answer.
allow_zero = random.random() < 0.2
# Repeat until we find a pair with an integral answer. (Avoids ambiguity with
# decimals.)
while True:
base_unit, target_unit = random.sample(list(dimension.keys()), 2)
base_value = number.non_integer_rational(2, signed=False)
if train_test_split.is_train(base_value) != is_train:
continue
answer = (base_value * sympy.Rational(dimension[base_unit]) /
sympy.Rational(dimension[target_unit]))
if (abs(answer) <= 100000 and sympy.denom(answer) == 1
and (allow_zero or answer != 0)):
break
template = random.choice([
'Ada berapa {target_name} di {base_value} dari sebuah {base_name}?',
'Berapa {base_value} dari {base_name} di {target_name}?',
])
if sympy.denom(base_value) > 20 or random.choice([False, True]):
base_value_string = base_value # Will be represented as e.g., 2/3.
else:
base_value_string = display.StringNumber(
base_value) # e.g., two thirds
question = example.question(context,
template,
base_name=base_unit.name,
base_value=base_value_string,
target_name=target_unit.name)
return example.Problem(question=question, answer=answer)