def get_stats(self, key):
"""
Lets an algorithm ask the logger for mean/std/min/max of a diagnostic.
"""
v = self.epoch_dict[key]
vals = np.concatenate(v) if isinstance(v[0], np.ndarray) and len(v[0].shape)>0 else v
return mpi_statistics_scalar(vals)
def log_tabular(self, key, val=None, with_min_and_max=False, average_only=False):
"""
Log a value or possibly the mean/std/min/max values of a diagnostic.
Args:
key (string): The name of the diagnostic. If you are logging a
diagnostic whose state has previously been saved with
``store``, the key here has to match the key you used there.
val: A value for the diagnostic. If you have previously saved
values for this key via ``store``, do *not* provide a ``val``
here.
with_min_and_max (bool): If true, log min and max values of the
diagnostic over the epoch.
average_only (bool): If true, do not log the standard deviation
of the diagnostic over the epoch.
"""
if val is not None:
super().log_tabular(key,val)
else:
v = self.epoch_dict[key]
vals = np.concatenate(v) if isinstance(v[0], np.ndarray) and len(v[0].shape)>0 else v
stats = mpi_statistics_scalar(vals, with_min_and_max=with_min_and_max)
super().log_tabular(key if average_only else 'Average' + key, stats[0])
if not(average_only):
super().log_tabular('Std'+key, stats[1])
if with_min_and_max:
super().log_tabular('Max'+key, stats[3])
super().log_tabular('Min'+key, stats[2])
self.epoch_dict[key] = []
def get(self):
assert self.ptr == self.max_size
self.ptr, self.path_start_idx = 0, 0
adv_mean, adv_std = mpi_statistics_scalar(self.adv_buf)
self.adv_buf = (self.adv_buf - adv_mean) / adv_std
return [
self.obs_buf, self.act_buf, self.adv_buf, self.ret_buf,
self.logp_buf
]
def get(self):
"""
Call this at the end of an epoch to get all of the data from
the buffer, with advantages appropriately normalized (shifted to have
mean zero and std one). Also, resets some pointers in the buffer.
"""
assert self.ptr == self.max_size # buffer has to be full before you can get
self.ptr, self.path_start_idx = 0, 0
# the next two lines implement the advantage normalization trick
adv_mean, adv_std = mpi_statistics_scalar(self.adv_buf)
self.adv_buf = (self.adv_buf - adv_mean) / adv_std
return [self.act_buf, self.adv_buf, self.ret_buf, self.logp_buf, self.rew_buf]
def get( self):
assert self.ptr == self.max_size
self.ptr, self.path_start_idx = 0,0
adv_mean, adv_std = mpi_statistics_scalar( self.adv_buf)
self.adv_buf = (self.adv_buf - adv_mean) / adv_std
def keys_as_sorted_list(dict):
return sorted(list(dict.keys()))
def values_as_sorted_list(dict):
return [dict[k] for k in keys_as_sorted_list(dict)]
return [self.obs_buf, self.act_buf, self.adv_buf, self.ret_buf, self.logp_buf] \
+ values_as_sorted_list(self.info_buffs)
def get(self):
"""
Call this at the end of an epoch to get all of the data from
the buffer, with advantages appropriately normalized (shifted to have
mean zero and std one). Also, resets some pointers in the buffer.
"""
assert self.ptr == self.max_size # buffer has to be full before you can get
self.ptr, self.path_start_idx = 0, 0
# the next two lines implement the advantage normalization trick
adv_mean, adv_std = mpi_statistics_scalar(self.adv_buf)
self.adv_buf = (self.adv_buf - adv_mean) / adv_std
data = dict(obs=self.obs_buf,
act=self.act_buf,
ret=self.ret_buf,
adv=self.adv_buf,
logp=self.logp_buf)
return {
k: torch.as_tensor(v, dtype=torch.float32)
for k, v in data.items()
}
