sled-umich/RACER-replay-public
收藏Hugging Face2024-10-15 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/sled-umich/RACER-replay-public
下载链接
链接失效反馈官方服务：
资源简介：
---
license: mit
---
Replay Buffers preprocessed by [YARR](https://github.com/stepjam/YARR). The statistics of transitions for training are as follows:
|Task| Number|
|--|--|
| close_jar | 6769 |
| insert_onto_square_peg | 5863 |
| light_bulb_in | 6654 |
| meat_off_grill | 5574 |
| open_drawer | 4281 |
| place_cups | 11112 |
| place_shape_in_shape_sorter | 4586 |
| place_wine_at_rack_location | 6160 |
| push_buttons | 4818 |
| put_groceries_in_cupboard | 4609 |
| put_item_in_drawer | 11802 |
| put_money_in_safe | 6194 |
| reach_and_drag | 5828 |
| slide_block_to_color_target | 3148 |
| stack_blocks | 16785 |
| stack_cups | 11301 |
| sweep_to_dustpan_of_size | 6072 |
| turn_tap | 2288 |
|Total|123844|
Unzip the files inside each directory like this:
```
racer_replay_public
├── close_jar
│ ├── 0.replay
│ ├── 1.replay
...
│ └── replay_info.npy
...
└── turn_tap
```
The total size of the replay buffers is around 2TB.
提供机构：
sled-umich
