4df11055e61fa6c9fede6b2114c8ce05de9a035e,slm_lab/agent/algorithm/dqn.py,DQNBase,compute_q_target_values,#DQNBase#Any#,57

Before Change

        // Make future reward 0 if the current state is done
        float_data_list = [
            "states", "actions", "rewards", "dones", "next_states"]
        for k in float_data_list:
            batch[k] = Variable(torch.from_numpy(batch[k]).float())
        // print("batch")
        // print(batch["states"])
        // print(batch["actions"])
        // print(batch["rewards"])
        // print(batch["dones"])
        // print(1 - batch["dones"])
        q_vals = self.net.wrap_eval(batch["states"])
        // print(f"q_vals {q_vals}")
        q_targets_all = batch["rewards"].data + self.gamma * \
            torch.mul((1 - batch["dones"].data),

After Change

        // Depending on the algorithm this is either the current
        // net or target net
        q_next_st_vals = self.eval_net.wrap_eval(batch["next_states"])
        idx = torch.from_numpy(np.array(list(range(self.batch_size))))
        q_next_st_vals_max = q_next_st_vals[idx, q_next_actions]
        q_next_st_vals_max.unsqueeze_(1)
        // Compute final q_target using reward and estimated
        // best Q value from the next state if there is one
        // Make future reward 0 if the current state is done

Instances