當前位置: 首頁>>代碼示例>>Python>>正文


Python rollout.RolloutWorker方法代碼示例

本文整理匯總了Python中baselines.her.rollout.RolloutWorker方法的典型用法代碼示例。如果您正苦於以下問題:Python rollout.RolloutWorker方法的具體用法?Python rollout.RolloutWorker怎麽用?Python rollout.RolloutWorker使用的例子?那麽, 這裏精選的方法代碼示例或許可以為您提供幫助。您也可以進一步了解該方法所在baselines.her.rollout的用法示例。


在下文中一共展示了rollout.RolloutWorker方法的2個代碼示例,這些例子默認根據受歡迎程度排序。您可以為喜歡或者感覺有用的代碼點讚,您的評價將有助於係統推薦出更棒的Python代碼示例。

示例1: main

# 需要導入模塊: from baselines.her import rollout [as 別名]
# 或者: from baselines.her.rollout import RolloutWorker [as 別名]
def main(policy_file, seed, n_test_rollouts, render):
    set_global_seeds(seed)

    # Load policy.
    with open(policy_file, 'rb') as f:
        policy = pickle.load(f)
    env_name = policy.info['env_name']

    # Prepare params.
    params = config.DEFAULT_PARAMS
    if env_name in config.DEFAULT_ENV_PARAMS:
        params.update(config.DEFAULT_ENV_PARAMS[env_name])  # merge env-specific parameters in
    params['env_name'] = env_name
    params = config.prepare_params(params)
    config.log_params(params, logger=logger)

    dims = config.configure_dims(params)

    eval_params = {
        'exploit': True,
        'use_target_net': params['test_with_polyak'],
        'compute_Q': True,
        'rollout_batch_size': 1,
        'render': bool(render),
    }

    for name in ['T', 'gamma', 'noise_eps', 'random_eps']:
        eval_params[name] = params[name]
    
    evaluator = RolloutWorker(params['make_env'], policy, dims, logger, **eval_params)
    evaluator.seed(seed)

    # Run evaluation.
    evaluator.clear_history()
    for _ in range(n_test_rollouts):
        evaluator.generate_rollouts()

    # record logs
    for key, val in evaluator.logs('test'):
        logger.record_tabular(key, np.mean(val))
    logger.dump_tabular() 
開發者ID:Hwhitetooth,項目名稱:lirpg,代碼行數:43,代碼來源:play.py

示例2: main

# 需要導入模塊: from baselines.her import rollout [as 別名]
# 或者: from baselines.her.rollout import RolloutWorker [as 別名]
def main(policy_file, seed, n_test_rollouts, render):
    set_global_seeds(seed)

    # Load policy.
    with open(policy_file, 'rb') as f:
        policy = pickle.load(f)
    env_name = policy.info['env_name']

    # Prepare params.
    params = config.DEFAULT_PARAMS
    if env_name in config.DEFAULT_ENV_PARAMS:
        params.update(config.DEFAULT_ENV_PARAMS[env_name])  # merge env-specific parameters in
    params['env_name'] = env_name
    params = config.prepare_params(params)
    config.log_params(params, logger=logger)

    dims = config.configure_dims(params)

    eval_params = {
        'exploit': True,
        'use_target_net': params['test_with_polyak'],
        'compute_Q': True,
        'rollout_batch_size': 1,
        'render': bool(render),
    }

    for name in ['T', 'gamma', 'noise_eps', 'random_eps']:
        eval_params[name] = params[name]

    evaluator = RolloutWorker(params['make_env'], policy, dims, logger, **eval_params)
    evaluator.seed(seed)

    # Run evaluation.
    evaluator.clear_history()
    for _ in range(n_test_rollouts):
        evaluator.generate_rollouts()

    # record logs
    for key, val in evaluator.logs('test'):
        logger.record_tabular(key, np.mean(val))
    logger.dump_tabular() 
開發者ID:hiwonjoon,項目名稱:ICML2019-TREX,代碼行數:43,代碼來源:play.py


注:本文中的baselines.her.rollout.RolloutWorker方法示例由純淨天空整理自Github/MSDocs等開源代碼及文檔管理平台,相關代碼片段篩選自各路編程大神貢獻的開源項目,源碼版權歸原作者所有,傳播和使用請參考對應項目的License;未經允許,請勿轉載。