9128a4b20336fb8c07fc4e0e0985fe299d1a5f12,tensorforce/core/memories/replay_memory.py,ReplayMemory,get_batch,#ReplayMemory#Any#,71

Before Change

        for n in range(len(self.internals)):
            batch["internals"].append(self.internals.take(indices))

        return batch

After Change

        else:
            indices = list(xrange(start, self.capacity)) + list(xrange(0, end))

        return dict(
            states={name: state.take(indices, axis=0) for name, state in self.states.items()},
            actions={name: action.take(indices) for name, action in self.actions.items()},
            rewards=self.rewards.take(indices),
            terminals=self.terminals.take(indices),
            internals=[internal.take(indices, axis=0) for internal in self.internals]
        )

Instances