task_id
stringclasses 5
values | suite
stringclasses 1
value | type
stringclasses 1
value | baseline_ppo_return
float64 -6.1
-3.2
|
|---|---|---|---|
place-in-drawer
|
calvin
|
manipulation
| -3.2
|
pick-up-block
|
calvin
|
manipulation
| -3.6
|
turn-on-lightbulb
|
calvin
|
manipulation
| -3.7
|
close-drawer
|
calvin
|
manipulation
| -4.2
|
open-drawer
|
calvin
|
manipulation
| -6.1
|
CALVIN Manipulation Benchmark
PPO evaluation benchmark on CALVIN tabletop manipulation tasks.
Tasks
| Task ID | Name |
|---|---|
| place-in-drawer | place_in_drawer |
| pick-up-block | pick_up_block |
| turn-on-lightbulb | turn_on_lightbulb |
| close-drawer | close_drawer |
| open-drawer | open_drawer |
Usage
Submit evaluation results to this benchmark by adding .eval_results/*.yaml files
to your model repo, referencing this dataset's task IDs.
漏 2026 ParamTatva.org
- Downloads last month
- 10