def _set_reward(self, reward_type, value):
"""Set the gain or loss to the given value."""
if self._got_reward:
raise ValueError("only one call to maximize or minimize is allowed")
if isinstance(value, np.ndarray):
if len(value.shape) != 1:
raise ValueError("gain/loss must be a scalar or 1-dimensional array, not {_coconut_format_0}".format(_coconut_format_0=(value)))
value = tuple(value)
self._current_example[reward_type] = denumpy_all(value)
if not self.is_serving:
self._save_current_data()
def _set_reward(self, reward_type, value):
"""Set the gain or loss to the given value."""
if self._got_reward:
raise ValueError("only one call to maximize or minimize is allowed")
if isinstance(value, np.ndarray):
if len(value.shape) != 1:
raise ValueError("gain/loss must be a scalar or 1-dimensional array, not {_coconut_format_0}".format(_coconut_format_0=(value)))
value = tuple(value)
self._current_example[reward_type] = denumpy_all(value)
if not self.is_serving:
self._save_current_data()
# _save_current_data ensures that _old_params has already been
# updated with _new_params, so _new_params can safely be cleared
self._new_params = {}
def standardize_args(self, func, args):
"""Standardize param func and args."""
# denumpy args
args = denumpy_all(args)
# detect invalid funcs
if func not in self.handlers:
raise ValueError(
"unknown parameter definition function {_coconut_format_0} (register with bbopt.params.param_processor.register)"
.format(_coconut_format_0=(func)))
# run handler
result = self.handlers[func](args)
args = result if result is not None else args
# standardize arguments to a list
return list(args)
def standardize_kwargs(self, kwargs):
"""Standardizes param keyword args."""
return fmap(lambda k, v: denumpy_all((k, v)), kwargs)