Spice + Cider

A generic RL environment to generate textual sequences. For eg: text generation, summarization, machine translation, text simplificat

Args: observation_space (DictSpace): Observation space action_space (Discrete): Action space lr_schedule

singular instance

Args: max_steps (int): max steps for each episode reward_function (RewardFunction): reward function that computes sca

Returns the probability that the table entry is mentioned in the sentence. A simple implementation which checks the longest common subsequence be

:param action_masks: Masks applied to constrain the choice of possible actions.

Adds annotated sample for sampling/replaying

Eliminate ("mask out") chosen distribution outcomes by setting their probability to 0. :param masks: An optional boolean ndarray of

Returns a dict where key is the metric name and value is again a dict consisting of tuple of individual scores (if any) and corpus level scor