1 個回答
- 最新
- 最多得票
- 最多評論
1
So for when using the console for training your AWS DeepRacer model you are limited to PPO and SAC as well as the hyperparameters and action space definitions, if you want to be able to tinker with more then you are better off looking at DeepRacer for Cloud which is tooling that has been built by the DeepRacer Community. If you're not already a member of this slack then I would recommend joining. Plenty of the top racers in the global league are active within the Slack and pretty generous with help and advice to help you improve your models. Good luck with your training and hopefully see you at the track at one of the global AWS Summits.
已回答 2 年前
相關內容
AWS 官方已更新 2 年前- AWS 官方已更新 2 年前
- AWS 官方已更新 2 年前
Thanks! I appreciate the response and I'm planning on joining the slack here soon. I have looked over DeepRacer for Cloud and Sagemaker and a few other Githubs and but everything seems to work relatively similarly when it comes to using Amazons Neural Net. I was hoping that I could actually make my Neural Net from scratch and select my own layers and weights and more than just the hyperparameters that are given.