Dataset Viewer
Auto-converted to Parquet Duplicate
task_id
stringclasses
10 values
suite
stringclasses
1 value
type
stringclasses
1 value
obs_dim
int64
39
39
act_dim
int64
4
4
max_steps
int64
150
150
baseline_ppo_return
float64
0.4
9.5
baseline_success_rate
float64
0
0.95
button-press-topdown-v3
metaworld
manipulation
39
4
150
6.1
0.95
drawer-close-v3
metaworld
manipulation
39
4
150
9.5
0.95
drawer-open-v3
metaworld
manipulation
39
4
150
7.7
0.95
window-open-v3
metaworld
manipulation
39
4
150
8
0.95
window-close-v3
metaworld
manipulation
39
4
150
5.2
0.95
reach-v3
metaworld
manipulation
39
4
150
8.3
0.25
door-open-v3
metaworld
manipulation
39
4
150
4.8
0.25
push-v3
metaworld
manipulation
39
4
150
2.7
0.1
peg-insert-side-v3
metaworld
manipulation
39
4
150
1.7
0
pick-place-v3
metaworld
manipulation
39
4
150
0.4
0.05

MetaWorld Manipulation Benchmark

PPO evaluation benchmark on Meta-World manipulation tasks.

Tasks

Task ID Name
button-press-topdown-v3 button-press-topdown-v3
drawer-close-v3 drawer-close-v3
drawer-open-v3 drawer-open-v3
window-open-v3 window-open-v3
window-close-v3 window-close-v3
reach-v3 reach-v3
door-open-v3 door-open-v3
push-v3 push-v3
peg-insert-side-v3 peg-insert-side-v3
pick-place-v3 pick-place-v3

Usage

Submit evaluation results to this benchmark by adding .eval_results/*.yaml files to your model repo, referencing this dataset's task IDs.

© 2026 ParamTatva.org

Downloads last month
12