This dataset was created using LeRobot.
Dataset Description
leapshared/sciedu_bench_v2_sim_props8_redox
SciEdu-Bench(과학 실험 시뮬 벤치) 학습 시연이다. 실험 하나(redox)의 순위 과제마다 시뮬레이터(MuJoCo)에서 지은 시연을 담았다. 평가는 꾸러미(config/bench/sciedu_bench_package.yaml)의 고정 시작 50개로 따로 한다. 이 데이터에는 평가 시작이 없다.
산화환원 옛 판(redox_*, props-8, 파란 용액 150 mL): 녹화 Redox_pauses_20261002(80편, 한 편에 실험 하나)의 단계 구간에서 지은 과제. 평가에 쓰지 않은 녹화가 과제마다 30편 안팎이고 그중 그 단계 구간이 있는 편은 몇 편뿐이라, 녹화 시작 하나에서 비커를 녹화 퍼짐 안에서 흔든 시작을 여러 개 뽑았다(평가 시작과 0.5 cm·5도 안이면 버림). 과제마다 편 수와 바탕 녹화 수를 아래에 적는다. 같은 녹화에서 뽑은 편끼리는 비커 자리만 다르다.
출처
- 장면 판
props-8, 규격 판2, 꾸러미 판0.7.0-draft(셋 중 하나라도 다르면 다른 데이터다). - 실험 현상 층을 켠 그림이다(산화환원 판의 구리 석출과 용액 색). 카메라는 물리 시계(1배, 규격
phenomena.observation_time_scale)로 그려 실물 녹화가 본 만큼만 보인다. 사람에게 보여 주는 가속 시계는 쓰지 않았다. 편 끝의 현상 값은sim_episodes.jsonl의phenomena에 있다. - 시작 자세의 출처 녹화:
leapshared/Redox_pauses_20261002_170818_subtask(로컬 녹화). - 생성 코드: 저장소 커밋
8f7d9c4879915bd012a7c2a519f8079af0af356f, 도구scripts/sim/bench_train_data.py, 짓는 법config/bench/sciedu_bench_train_data.yaml.
생성기
녹화를 재생해 만든 편은 없다(녹화 재생은 학습 데이터 출처가 아니다, 2026-11-06). 편마다 만든 법:
- 녹화 손 길을 흔든 물체에 실은 IK 코드: 515편
판정(평가와 같은 술어)을 통과한 편만 담았다. 행동은 생성기가 보낸 그대로이고, 같은 시작에서 같은 행동을 카메라를 켜고 다시 돌려 그림을 넣었다(판정과 길이가 같은 것만).
과제와 편수
| 과제 | 묶음 | 문장(task) |
편/목표 | 만든 법 | 프레임 |
|---|---|---|---|---|---|
redox_lift |
spatial | lift the metal plate out of the left beaker | 47/50 | 녹화 손 길을 흔든 물체에 실은 IK 코드 47 | 10158 |
redox_move_blue |
spatial | move the plate over the blue solution | 50/50 | 녹화 손 길을 흔든 물체에 실은 IK 코드 50 | 6320 |
redox_put_left |
spatial | put the plate into the left beaker | 50/50 | 녹화 손 길을 흔든 물체에 실은 IK 코드 50 | 7970 |
redox_move_blue_look |
object | move the plate over the blue solution | 50/50 | 녹화 손 길을 흔든 물체에 실은 IK 코드 50 | 6537 |
redox_point |
goal | point at the plate with the left hand | 50/50 | 녹화 손 길을 흔든 물체에 실은 IK 코드 50 | 8848 |
redox_dip |
goal | lower the plate into the blue solution | 50/50 | 녹화 손 길을 흔든 물체에 실은 IK 코드 50 | 9475 |
redox_move_left |
goal | move the plate over the left beaker | 50/50 | 녹화 손 길을 흔든 물체에 실은 IK 코드 50 | 7298 |
redox_script |
long | lift the metal plate out of the left beaker | 18/50 | 녹화 손 길을 흔든 물체에 실은 IK 코드 18 | 70498 |
redox_raise |
rest | raise the plate out of the solution | 50/50 | 녹화 손 길을 흔든 물체에 실은 IK 코드 50 | 5023 |
redox_point_again |
rest | point at the dipped plate with the left hand | 50/50 | 녹화 손 길을 흔든 물체에 실은 IK 코드 50 | 8612 |
redox_home |
rest | return to the home posture | 50/50 | 녹화 손 길을 흔든 물체에 실은 IK 코드 50 | 5743 |
합 515편, 146482프레임, 30 fps. 못 채운 과제: redox_lift 47/50, redox_script 18/50. 시작을 뽑은 녹화 에피소드 수: redox_lift 17, redox_move_blue 9, redox_put_left 5, redox_move_blue_look 9, redox_point 5, redox_dip 8, redox_move_left 3, redox_script 1, redox_raise 7, redox_point_again 9, redox_home 14.
평가 시작과의 분리
- 규칙: 과제마다 그 과제의 평가 시작이 쓰는 녹화 에피소드를 뺀 나머지(같은 실험의 다른 과제와는 겹칠 수 있다, 점검이 센다)
- 잰 것(
scripts/sim/bench_train_data.py check): 같은 과제의 평가 시작과 녹화 에피소드가 겹치는 편 0, 평가 시작과 같은 장면(소품 0.5 cm·팔 5.0도 안) 0, 같은 실험 다른 과제의 평가 시작과 녹화 에피소드를 나누는 편 440(같은 녹화의 다른 단계). - 학습 시작에서 가장 가까운 평가 시작까지 물체 x·y 거리:
redox_lift가장 가까운 것 0.82 cm·가운데 2.8 cm;redox_move_blue가장 가까운 것 1.19 cm·가운데 3.0 cm;redox_put_left가장 가까운 것 0.99 cm·가운데 1.85 cm;redox_move_blue_look가장 가까운 것 0.8 cm·가운데 2.85 cm;redox_point가장 가까운 것 1.03 cm·가운데 2.21 cm;redox_dip가장 가까운 것 0.84 cm·가운데 2.38 cm;redox_move_left가장 가까운 것 1.15 cm·가운데 1.95 cm;redox_raise가장 가까운 것 1.04 cm·가운데 2.2 cm;redox_point_again가장 가까운 것 0.65 cm·가운데 2.42 cm;redox_home가장 가까운 것 0.98 cm·가운데 3.22 cm;redox_script가장 가까운 것 1.57 cm·가운데 1.91 cm.
꼴
실물 녹화(LeRobot v3.0)와 같다: 30 fps, action·observation.state 16값(팔로워 관절 도, 왼팔 7·왼 그리퍼·오른팔 7·오른 그리퍼, 절대 관절값), 카메라 셋(observation.images.follower_d455f 가슴, left_wrist, right_wrist, 480×640, AV1). robot_type은 openarm_sciedu_sim(꾸러미 io.robot_type_sim)이라 실험 데이터셋끼리 LeRobot 합치기로 하나가 된다. task는 평가가 주는 규격 문장 그대로다(머리말 없음). 실물 녹화의 문장은 앞에 "Do the ... experiment. "가 붙는다. 녹화에 없는 규격 문장: point at the dipped plate with the left hand(라벨 파일 기준).
에피소드마다 과제·만든 법·시작(녹화 에피소드·흔들기·소품 자리)·판정은 sim_episodes.jsonl에 있다.
- 끝 멈춤(행동이 마지막 값에 머문 시간) 가운데: 시연 0.47~2.57초, 출처 녹화 0.42초. 판정을 통과한 틱 뒤에 마지막 명령으로 0.43초(출처 녹화 끝 멈춤의 가운데값) 더 서 있는 틱을 담았다. 그동안 물건이 1 cm 넘게 가거나 5도 넘게 돌거나 쥔 것을 놓친 편은 버렸다(
sim_episodes.jsonl의end_hold). - 이끎 구간(시작 자세에서 과제 첫 자세로 가는 틱,
lead_ticks)이 있는 편 104개. 녹화가 아니므로 거를 수 있다.
라이선스
apache-2.0(허브 메타데이터 apache-2.0, 선생님 2026-11-11, 실물 녹화 leapshared/SciEdu_mineral_steps_20261109와 같게). 이 시뮬 시연 데이터셋의 라이선스다. 코드·장면(소품 CAD·사진 결)의 라이선스는 꾸러미의 license 칸에서 따로 정한다.
판(리비전) 고정
학습·평가 기록에는 이 데이터를 저장소@커밋으로 적는다(꾸러미 training_data.sim이 커밋을 고정한다). 부를 때도 커밋을 준다:
lerobot-train --dataset.repo_id=leapshared/sciedu_bench_v2_sim_props8_redox --dataset.revision=<커밋> ...
from lerobot.datasets.lerobot_dataset import LeRobotDataset
ds = LeRobotDataset("leapshared/sciedu_bench_v2_sim_props8_redox", revision="<커밋>")
장면·판정·시작이 바뀌면 같은 저장소에 덮어쓰지 않고 장면 판이 이름에 든 새 저장소로 올린다. 다시 짓기: scripts/sim/bench_train_data.py all --dataset redox.
Training demonstrations of SciEdu-Bench (MuJoCo), experiment redox, scene props-8, spec 2, package 0.7.0-draft. 515 episodes made in simulation by recorded_path_shifted (515); no recording replay. Only judged successes are kept. The training starts are disjoint from the package's fixed evaluation starts (same-task recorded-episode overlap 0, starts within 0.5 cm and 5.0 degrees of one 0). Schema: the real recordings' (30 fps, 16 absolute joint degrees, three 480x640 AV1 cameras); task is the benchmark sentence without the recordings' head. License: apache-2.0 (metadata apache-2.0). Drawn with the phenomenon layer on (redox) on the physical clock (1x, spec phenomena.observation_time_scale): the cameras see what the real recordings saw, not the sped-up show clock; each episode's end values are in sim_episodes.jsonl phenomena. Pin a commit (--dataset.revision). Old redox board (redox_*, props-8, 150 mL blue solution) from the step spans of the 80 whole-experiment recordings. Only a few recordings outside the eval ones hold a task's span, so each gives several demo starts with the beakers shifted within the recording's spread (none within 0.5 cm and 5 degrees of an eval start); the counts of demos and of source recordings per task are below. Demos from one recording differ only in where the beakers stand.
- Homepage: [More Information Needed]
- Paper: [More Information Needed]
- License: apache-2.0
Dataset Structure
{
"codebase_version": "v3.0",
"fps": 30,
"features": {
"action": {
"dtype": "float32",
"shape": [
16
],
"names": [
"left_joint_1.pos",
"left_joint_2.pos",
"left_joint_3.pos",
"left_joint_4.pos",
"left_joint_5.pos",
"left_joint_6.pos",
"left_joint_7.pos",
"left_gripper.pos",
"right_joint_1.pos",
"right_joint_2.pos",
"right_joint_3.pos",
"right_joint_4.pos",
"right_joint_5.pos",
"right_joint_6.pos",
"right_joint_7.pos",
"right_gripper.pos"
]
},
"observation.state": {
"dtype": "float32",
"shape": [
16
],
"names": [
"left_joint_1.pos",
"left_joint_2.pos",
"left_joint_3.pos",
"left_joint_4.pos",
"left_joint_5.pos",
"left_joint_6.pos",
"left_joint_7.pos",
"left_gripper.pos",
"right_joint_1.pos",
"right_joint_2.pos",
"right_joint_3.pos",
"right_joint_4.pos",
"right_joint_5.pos",
"right_joint_6.pos",
"right_joint_7.pos",
"right_gripper.pos"
]
},
"observation.images.follower_d455f": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height": 480,
"video.width": 640,
"video.codec": "av1",
"video.pix_fmt": "yuv420p",
"video.fps": 30,
"video.channels": 3,
"has_audio": false,
"video.g": 2,
"video.crf": 30,
"video.preset": 12,
"video.fast_decode": 0,
"video.video_backend": "pyav",
"video.extra_options": {},
"is_depth_map": false
}
},
"observation.images.left_wrist": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height": 480,
"video.width": 640,
"video.codec": "av1",
"video.pix_fmt": "yuv420p",
"video.fps": 30,
"video.channels": 3,
"has_audio": false,
"video.g": 2,
"video.crf": 30,
"video.preset": 12,
"video.fast_decode": 0,
"video.video_backend": "pyav",
"video.extra_options": {},
"is_depth_map": false
}
},
"observation.images.right_wrist": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height": 480,
"video.width": 640,
"video.codec": "av1",
"video.pix_fmt": "yuv420p",
"video.fps": 30,
"video.channels": 3,
"has_audio": false,
"video.g": 2,
"video.crf": 30,
"video.preset": 12,
"video.fast_decode": 0,
"video.video_backend": "pyav",
"video.extra_options": {},
"is_depth_map": false
}
},
"timestamp": {
"dtype": "float32",
"shape": [
1
],
"names": null
},
"frame_index": {
"dtype": "int64",
"shape": [
1
],
"names": null
},
"episode_index": {
"dtype": "int64",
"shape": [
1
],
"names": null
},
"index": {
"dtype": "int64",
"shape": [
1
],
"names": null
},
"task_index": {
"dtype": "int64",
"shape": [
1
],
"names": null
}
},
"total_episodes": 515,
"total_frames": 146482,
"total_tasks": 10,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 200,
"data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
"video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4",
"robot_type": "openarm_sciedu_sim",
"splits": {
"train": "0:515"
}
}
Citation
BibTeX:
[More Information Needed]
- Downloads last month
- 135