Downloads · 30 days
0
guangmokuai
yangzhaofeng/guangmokuai
guangmokuai is a reinforcement learning model from yangzhaofeng. Use it for the reinforcement learning task on the model card, and read the license before you ship it in a product. It is set up for lerobot. The card lists the license as apache-2.0.
本模型集合收录了基于 LeRobot 0.4.4 训练的多个 ACT 策略模型。所有模型均训练 100,000 步，支持多相机视觉输入和触觉传感器数据。训练所用数据集包含千觉传感器和戴盟传感器的数据集集合。
Access
Public
Updated Aug 11, 2026
Repo size
68.2 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors34.5 GB · 98%
At a glance
- Task
- Reinforcement Learning
- Library
- lerobot
- License
- apache-2.0
- Access
- Public
- Created
- Jul 27, 2026
- Updated
- Aug 11, 2026
- SHA
- 7d882574
Datasets
Model card
From the Hugging Face model README
LeRobot 触觉增强机械臂策略模型集合
本模型集合收录了基于 LeRobot 0.4.4 训练的多个 ACT 策略模型。所有模型均训练 100,000 步，支持多相机视觉输入和触觉传感器数据。训练所用数据集包含千觉传感器和戴盟传感器的数据集集合。
千觉 传感器
| 模型位置 | policy | 数据集 | 版本 | 训练步数 | 输入 | 成功率 |
|---|---|---|---|---|---|---|
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/qj/old/qj_100" style="color:red">qj_100<br>(desperate)</a> | 共享ACT | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/qj/old/qj_100" style="color:red">qj_100<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br>+robot state | 9% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/qj/old/qj_150(100%2B50)" style="color:red">qj_150<br>(desperate)</a> | 共享ACT | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/qj/old" style="color:red">qj_jiaozheng_50+qj_100<br>(desperate)</a> | 0.4.4 | 100k | 3 相机<br>+robot state | 18% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/qj/old/qj_200(100%2B100)" style="color:red">qj_200<br>(desperate)</a> | 共享ACT | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/qj/old/qj_200(100%2B100)" style="color:red">qj_200(100+100)<br>(desperate)</a> | 0.4.4 | 100k | 3 相机<br>+robot state | 6% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/qj/old/qj_100_reinforce" style="color:red">qj_100_reinforce<br>(desperate)</a> | 共享ACT+图像增强 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/qj/old/qj_100%20(1)" style="color:red">qj_100（1）<br>（desperate）</a> | 0.4.4 | 100k | 3 相机 <br> + robot state | 42% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/qj/old/qj_300_reinforce" style="color:red">qj_300_reinforce<br>(desperate)</a> | 共享ACT+图像增强 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/qj/old/qj_300" style="color:red">qj_300<br>（desperate）</a> | 0.4.4 | 100k | 3 相机 <br> + robot state | 60% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/qj/old/qj_50_reinforce" style="color:red">qj_50_reinforce<br>(desperate)</a> | 共享ACT+图像增强 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/qj/old/qj_50" style="color:red">qj_50<br>（desperate）</a> | 0.4.4 | 100k | 3 相机 <br> + robot state | --- |
戴盟 传感器
| 模型位置 | policy | 数据集 | 版本 | 训练步数 | 输入 | 成功率 |
|---|---|---|---|---|---|---|
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/ACT/dm_act_100" style="color:red">dm_act_100<br>(desperate)</a> | 独立ACT | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_100" style="color:red">dm_100<br>（desperate）</a> | 0.4.4 | 100k | 3 相机 <br>+ distforce <br> + robot state | 17% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/ACT/dm_act_tac_100" style="color:red">dm_act_tac_100<br>(desperate)</a> | 独立ACT+触觉 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_100" style="color:red">dm_100<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + distforce <br> + robot state | --- |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/ACT_orin/dm_act_orin_100" style="color:red">dm_act_orin_100<br>(desperate)</a> | 共享ACT | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_100" style="color:red">dm_100<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + robot state | 42% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/ACT_orin/tac/dm_act_orin_tac_100" style="color:red">dm_act_orin_tac_100<br>(desperate)</a> | 共享ACT+触觉 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_100" style="color:red">dm_100<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + distforce <br> + robot state | 30% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/ACT/dm_act_200" style="color:red">dm_act_200<br>(desperate)</a> | 独立ACT | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_200" style="color:red">dm_200<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + distforce <br> + robot state | --- |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/ACT/dm_act_tac_200" style="color:red">dm_act_tac_200<br>(desperate)</a> | 独立ACT+触觉 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_200" style="color:red">dm_200<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + distforce <br> + robot state | --- |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/ACT_orin/dm_act_orin_200" style="color:red">dm_act_orin_200<br>(desperate)</a> | 共享ACT | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_200" style="color:red">dm_200<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + robot state | 26% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/ACT_orin/tac/dm_act_orin_tac_200" style="color:red">dm_act_orin_tac_200<br>(desperate)</a> | 共享ACT+触觉 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_200" style="color:red">dm_200<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> +distforce <br> + robot state | --- |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/ACT_orin/dm_act_orin_350" style="color:red">dm_act_orin_350<br>(desperate)</a> | 共享ACT | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_350" style="color:red">dm_350<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + robot state | 62% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/ACT_orin/reinforce/dm_act_orin_350_reinforce" style="color:red">dm_act_orin_350<br>reinforce(desperate)</a> | 共享ACT+图像增强 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_350" style="color:red">dm_350<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + robot state | 83% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/ACT_orin/dm_act_orin_500" style="color:red">dm_act_orin_500<br>(desperate)</a> | 共享ACT | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_500" style="color:red">dm_500<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + robot state | 15% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/ACT_orin/reinforce/dm_act_orin_500_reinforce" style="color:red">dm_act_orin_500<br>reinforce(desperate)</a> | 共享ACT+图像增强 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_500" style="color:red">dm_500<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + robot state | 26% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_act_orin_200_reinforce" style="color:red">dm_act_orin_200<br>reinforce(desperate)</a> | 共享ACT+图像增强 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_200%20(1)" style="color:red">dm_200(1)<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + robot state | 66% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_act_orin_350_reinforce" style="color:red">dm_act_orin_350<br>reinforce(desperate)</a> | 共享ACT+图像增强 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_350%20(1)" style="color:red">dm_350(1)<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + robot state | 64% |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_act_orin_350+50_reinforce_resume" style="color:red">dm_act_orin_350+50<br>reinforce_resume<br>(desperate)</a> | 共享ACT+图像增强 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_eval_rg_50" style="color:red">dm_eval_rg_50<br>（desperate）</a> | 0.4.4 | <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_act_orin_350_reinforce" style="color:red">dm_act_orin_350<br>reinforce(desperate)</a> <br>上resume训练100k | 3 相机<br> + robot state | --- |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_act_orin_350%2B50_reinforce" style="color:red">dm_act_orin_350+50<br>reinforce(desperate)</a> | 共享ACT+图像增强 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_350%2B50" style="color:red">dm_350+50<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + robot state | --- |
| <a href="https://huggingface.co/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_act_orin_tac_350%2B50_reinforce" style="color:red">dm_act_orin_tac<br>350+50_reinforce<br>(desperate)</a> | 共享ACT+图像增强 | <a href="https://huggingface.co/datasets/yangzhaofeng/guangmokuai/tree/main/dm/old/dm_350%2B50" style="color:red">dm_350+50<br>（desperate）</a> | 0.4.4 | 100k | 3 相机<br> + distforce <br> + robot state | --- |
| dm_act_orin_200<br>reinforce | 共享ACT+图像增强 | dm_200 | 0.4.4 | 100k | 3 相机<br> + robot state | 66% |
| dm_act_orin_350<br>reinforce | 共享ACT+图像增强 | dm_350 | 0.4.4 | 100k | 3 相机<br> + robot state | 76% |
| dm_act_orin_tac<br>distf_resnet_350<br>reinforce | 共享ACT+触觉+图像增强 | dm_350 | 0.4.4 | 100k | 3 相机<br> + distforce <br> +robot state | 42% |
模型详情
- 开发者: yangzhaofeng
- 模型架构: ACT(Action Chunking with Transformers)
- 框架版本: LeRobot 0.4.4
- 许可证: Apache 2.0
- Vision Backbone: ResNet18
文件结构
每个模型检查点包含：
model_name/
├── checkpoints/
│ ├── pretrained_model/
│ │ ├── config.json # 模型配置
│ │ ├── model.safetensors # 模型权重
│ │ ├── policy_preprocessor.json # 预处理器配置
│ │ ├── policy_preprocessor_step_3_normalizer_processor.safetensors
│ │ ├── policy_postprocessor.json
│ │ ├── policy_postprocessor_step_0_unnormalizer_processor.safetensors
│ │ └── train_config.json # 训练配置
│ └── training_state/
│ ├── optimizer_state.safetensors
│ ├── optimizer_param_groups.json
│ ├── rng_state.safetensors
│ ├── scheduler_state.json # (部分模型)
│ └── training_step.json
└── wandb/ # 训练日志
- Task
- Reinforcement Learning
- Library
- lerobot
- License
- apache-2.0
- Languages
- zh
- Created
- Jul 27, 2026
- Updated
- Aug 11, 2026