def test_a_negated_tag_is_also_interesting(rnd):
selector = TargetSelector(rnd)
selector.add(FakeConjectureData(tags=frozenset({0})))
selector.add(FakeConjectureData(tags=frozenset({0})))
selector.add(FakeConjectureData(tags=frozenset()))
_, data = selector.select()
assert not data.tags
def test_cycles_through_all_tags_in_bounded_time_mixed(rnd, d1, d2):
selector = TargetSelector(rnd)
for d in d1:
selector.add(d)
check_bounded_cycle(selector)
for d in d2:
selector.add(d)
check_bounded_cycle(selector)
def test_always_starts_with_rare_tags(rnd, n):
selector = TargetSelector(rnd)
selector.add(FakeConjectureData(tags=frozenset({0})))
for _ in hrange(n):
selector.select()
selector.add(FakeConjectureData(tags=frozenset({1})))
_, data = selector.select()
assert 1 in data.tags
def test_target_selector_will_use_novel_examples_preferentially():
selector = TargetSelector(random=Random(0), pool_size=3)
seen = set()
for i in range(100):
selector.add(FakeData())
assert len(selector) == min(i + 1, 3)
t = selector.select().global_identifier
assert t not in seen
seen.add(t)
def test_selects_non_universal_tag(rnd):
selector = TargetSelector(rnd)
selector.add(FakeConjectureData({0}))
selector.add(FakeConjectureData(set()))
tag1, x = selector.select()
assert tag1 is not universal
tag2, y = selector.select()
assert tag2 is not universal
assert tag1 != tag2
assert x != y
def test_target_selector_will_maintain_a_bounded_size_with_scores(pool_size):
selector = TargetSelector(random=Random(0), pool_size=pool_size)
selector.add(FakeData())
for i in range(100):
selector.add(
FakeData(target_observations={str(i // 3 == 0): float(i % 30)}))
assert len(selector) <= pool_size
for label, pool in selector.scored_examples.items():
scores = [ex.target_observations[label] for ex in pool]
assert scores == sorted(scores, reverse=True)
selector.add(FakeData())
assert len(selector) <= pool_size
def test_target_selector_will_eventually_reuse_examples():
selector = TargetSelector(random=Random(0), pool_size=2)
seen = set()
selector.add(FakeData())
selector.add(FakeData())
for _ in range(2):
x = selector.select()
assert x.global_identifier not in seen
seen.add(x.global_identifier)
for _ in range(2):
x = selector.select()
assert x.global_identifier in seen
def test_target_selector_will_maintain_a_bounded_pool():
selector = TargetSelector(random=Random(0), pool_size=3)
for i in range(100):
selector.add(FakeData())
assert len(selector) == min(i + 1, 3)
def test_cycles_through_all_tags_in_bounded_time(rnd, datas):
selector = TargetSelector(rnd)
for d in datas:
selector.add(d)
check_bounded_cycle(selector)
def test_target_selector_can_discard_labels():
selector = TargetSelector(random=Random(0), pool_size=2)
for i in range(10):
selector.add(FakeData(target_observations={str(i): 0.0}))
assert len(selector) <= 2
def initialize(self, rnd):
self.target_selector = TargetSelector(rnd)