From ae5cfe2f64ef6163a6006d6e470c2cdb0c433928 Mon Sep 17 00:00:00 2001 From: Kees Jan Koster Date: Mon, 24 Jul 2023 21:00:05 +0100 Subject: [PATCH] Chase interface changes to OpenAI's Gym package. --- .../DDPG/pytorch/lunar-lander/main_torch.py | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/ReinforcementLearning/PolicyGradient/DDPG/pytorch/lunar-lander/main_torch.py b/ReinforcementLearning/PolicyGradient/DDPG/pytorch/lunar-lander/main_torch.py index bba7c0f..04ce379 100644 --- a/ReinforcementLearning/PolicyGradient/DDPG/pytorch/lunar-lander/main_torch.py +++ b/ReinforcementLearning/PolicyGradient/DDPG/pytorch/lunar-lander/main_torch.py @@ -12,13 +12,14 @@ score_history = [] for i in range(1000): - obs = env.reset() + obs, info = env.reset() done = False + truncated = False score = 0 - while not done: + while not (done or truncated): act = agent.choose_action(obs) - new_state, reward, done, info = env.step(act) - agent.remember(obs, act, reward, new_state, int(done)) + new_state, reward, done, truncated, info = env.step(act) + agent.remember(obs, act, reward, new_state, int(done or truncated)) agent.learn() score += reward obs = new_state