Refactor code structure for improved readability and maintainability
This commit is contained in:
@@ -16,4 +16,4 @@ batch_size: 256 # 每次梯度更新抽样的 batch 大小
|
||||
# --- 训练循环控制 ---
|
||||
max_episodes: 200 # 总共训练多少个 episode
|
||||
max_steps: 200 # 每 episode 最多步数 (Pendulum-v1 默认 200 步截断)
|
||||
start_steps: 10000 # 纯随机动作探索的步数
|
||||
start_steps: 3000 # 纯随机动作探索的步数
|
||||
Reference in New Issue
Block a user