ONNX 模型库
返回模型

说明文档

PPO 智能体游玩 SnowballTarget

这是一个使用 Unity ML-Agents 库 训练的 PPO 智能体游玩 SnowballTarget 的模型。

使用方法（配合 ML-Agents）

文档地址：https://unity-technologies.github.io/ml-agents/ML-Agents-Toolkit-Documentation/

我们编写了完整的教程，帮助你学习如何使用 ML-Agents 训练你的第一个智能体并将其发布到 Hub：

  • 一个简短教程，教你让小狗 Huggy 🐶 去捡棍子，然后直接在浏览器中与它互动： https://huggingface.co/learn/deep-rl-course/unitbonus1/introduction
  • 一个更详细的教程，帮助你理解 ML-Agents 的工作原理： https://huggingface.co/learn/deep-rl-course/unit5/introduction

继续训练

mlagents-learn <your_configuration_file_path.yaml> --run-id=<run_id> --resume

观看你的智能体游玩

你可以直接在浏览器中观看你的智能体游玩

  1. 如果环境属于 ML-Agents 官方环境，请访问 https://huggingface.co/unity
  2. 第一步：找到你的 model_id：DhruvAwasthi/ppo-SnowballTarget
  3. 第二步：选择你的 .nn /.onnx 文件
  4. 点击"观看智能体游玩"👀

DhruvAwasthi/ppo-SnowballTarget

作者 DhruvAwasthi

reinforcement-learning ml-agents
↓ 1 ♥ 0

创建时间: 2023-06-09 22:15:36+00:00

更新时间: 2023-06-09 22:15:39+00:00

在 Hugging Face 上查看

文件 (30)

.gitattributes
README.md
SnowballTarget.onnx ONNX
SnowballTarget/SnowballTarget-149984.onnx ONNX
SnowballTarget/SnowballTarget-149984.pt
SnowballTarget/SnowballTarget-199984.onnx ONNX
SnowballTarget/SnowballTarget-199984.pt
SnowballTarget/SnowballTarget-249944.onnx ONNX
SnowballTarget/SnowballTarget-249944.pt
SnowballTarget/SnowballTarget-299968.onnx ONNX
SnowballTarget/SnowballTarget-299968.pt
SnowballTarget/SnowballTarget-349992.onnx ONNX
SnowballTarget/SnowballTarget-349992.pt
SnowballTarget/SnowballTarget-399992.onnx ONNX
SnowballTarget/SnowballTarget-399992.pt
SnowballTarget/SnowballTarget-449952.onnx ONNX
SnowballTarget/SnowballTarget-449952.pt
SnowballTarget/SnowballTarget-499976.onnx ONNX
SnowballTarget/SnowballTarget-499976.pt
SnowballTarget/SnowballTarget-500104.onnx ONNX
SnowballTarget/SnowballTarget-500104.pt
SnowballTarget/SnowballTarget-99960.onnx ONNX
SnowballTarget/SnowballTarget-99960.pt
SnowballTarget/checkpoint.pt
SnowballTarget/events.out.tfevents.1686347574.91b234421603.2573.0
config.json
configuration.yaml
run_logs/Player-0.log
run_logs/timers.json
run_logs/training_status.json