def write_stats(self, category: str, key: str, value: float,
step: int) -> None:
self._maybe_create_summary_writer(category)
summary = tf.Summary()
summary.value.add(tag="{}".format(key), simple_value=value)
self.summary_writers[category].add_summary(summary, step)
self.summary_writers[category].flush()
def write_stats(self, category: str, values: Dict[str, StatsSummary],
step: int) -> None:
self._maybe_create_summary_writer(category)
for key, value in values.items():
summary = tf.Summary()
summary.value.add(tag="{}".format(key), simple_value=value.mean)
self.summary_writers[category].add_summary(summary, step)
self.summary_writers[category].flush()
def write_summary(
self, global_step: int, delta_train_start: float, lesson_num: int = 0
) -> None:
"""
Saves training statistics to Tensorboard.
:param delta_train_start: Time elapsed since training started.
:param lesson_num: Current lesson number in curriculum.
:param global_step: The number of steps the simulation has been going for
"""
if (
global_step % self.trainer_parameters["summary_freq"] == 0
and global_step != 0
):
is_training = (
"Training."
if self.is_training and self.get_step <= self.get_max_steps
else "Not Training."
)
step = min(self.get_step, self.get_max_steps)
if len(self.stats["Environment/Cumulative Reward"]) > 0:
mean_reward = np.mean(self.stats["Environment/Cumulative Reward"])
LOGGER.info(
" {}: {}: Step: {}. "
"Time Elapsed: {:0.3f} s "
"Mean "
"Reward: {:0.3f}"
". Std of Reward: {:0.3f}. {}".format(
self.run_id,
self.brain_name,
step,
delta_train_start,
mean_reward,
np.std(self.stats["Environment/Cumulative Reward"]),
is_training,
)
)
set_gauge(f"{self.brain_name}.mean_reward", mean_reward)
else:
LOGGER.info(
" {}: {}: Step: {}. No episode was completed since last summary. {}".format(
self.run_id, self.brain_name, step, is_training
)
)
summary = tf.Summary()
for key in self.stats:
if len(self.stats[key]) > 0:
stat_mean = float(np.mean(self.stats[key]))
summary.value.add(tag="{}".format(key), simple_value=stat_mean)
self.stats[key] = []
summary.value.add(tag="Environment/Lesson", simple_value=lesson_num)
self.summary_writer.add_summary(summary, step)
self.summary_writer.flush()