Serendipity/rl-atari
1
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: 添加并行训练脚本和奖励塑形以改进PPO性能

Browse Source 
引入并行环境训练脚本 train_parallel_improved.py，实现多进程并行数据收集
添加奖励塑形包装器，根据速度、赛道位置和完成圈数调整奖励信号
优化神经网络结构和训练参数，包括更大的rollout缓冲区
删除旧的tensorboard日志文件，创建新的训练运行记录
This commit is contained in:
刘航宇
2026-05-01 09:26:39 +08:00
parent d6860f1f15
commit 7dea00195e
12 changed files with 662 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files