b5e848af46b4a6fc21e718803dbf3d7c72afd72a,ch08/lib/validation.py,,validation_run,#,9

Before Change

                obs_v = obs_v.cuda()
            out_v = net(obs_v)

            action_idx = out_v.max(dim=1)[1].data.cpu().numpy()[0]
            if np.random.random() < epsilon:
                action_idx = env.action_space.sample()
            action = environ.Actions(action_idx)

After Change

        episode_steps = 0

        while True:
            obs_v = torch.tensor([obs]).to(device)
            out_v = net(obs_v)

            action_idx = out_v.max(dim=1)[1].item()
            if np.random.random() < epsilon:

Instances