def save_mock_data(cache: SummaryCache):
i = 1
for j in range(500):
cache.store(reward=(1 - np.exp(i)) + 0.1 * np.random.rand(),
activations=np.random.normal(5, 0.5),
actions=np.random.normal(50, 0.5),
scalar=np.random.rand(1)[0],
uneven_length=np.array([1] * np.random.randint(0, 5)))
def test_single_key():
print_helper = PrintHelper()
cache = SummaryCache(mode='tiled')
print()
for i in range(10):
save_mock_data(cache)
mean_reward = cache.get("reward").mean()
assert isinstance(mean_reward, float)
def save_mock_data(cache: SummaryCache):
i = 1
for j in range(500):
cache.store(
reward=(1 - np.exp(i)) + 0.1 * np.random.rand(),
activations=np.random.normal(5, 0.5),
actions=np.random.normal(50, 0.5),
)
def test_summary_cache():
print_helper = PrintHelper()
cache = SummaryCache(mode='tiled')
print()
for i in range(10):
save_mock_data(cache)
stats = cache.summary_stats(reward="mean",
activations="quantile",
actions="histogram")
s = print_helper.format_tabular(stats)
def test_repr():
print_helper = PrintHelper()
cache = SummaryCache(mode='tiled')
print()
for i in range(10):
save_mock_data(cache)
stats = cache.summary_stats(reward="mean",
activations="quantile",
actions="histogram")
# s = print_helper.format_tabular(stats)
# s = print_helper.format_row_table(stats)
# print(s)
np.mean(cache.get('reward'))
from textwrap import dedent
assert str(cache) == dedent("""
SummaryCache: reward: Shape()[:5000]
activations: [5.370795870404581, 4.241508632737578][:5000]
actions: [50.77645686095641, 49.56686189467056][:5000]
scalar: Shape()[:5000]
uneven_length: Shape(1,)[:5000]
""")[1:]