Hugging Face's logo

Datasets:
sqOasis
/
rmbench-mem0-evaluation

Modalities:
Text
Video
Formats:
text
Size:
< 1K
Tags:
robotics
rmbench
evaluation
Libraries:
Datasets
License:
Dataset card Data Studio Files
xet
 Community
Dataset Viewer
Auto-converted to Parquet Duplicate
text
stringlengths
0
101
Eval log; authoritative episode records: episodes.json
episode_id=0, seed=100004, result=Fail
episode_id=1, seed=100009, result=Fail
episode_id=2, seed=100015, result=Success
episode_id=3, seed=100017, result=Fail
episode_id=4, seed=100022, result=Fail
episode_id=5, seed=100026, result=Fail
episode_id=6, seed=100027, result=Fail
episode_id=7, seed=100030, result=Fail
episode_id=8, seed=100033, result=Fail
episode_id=9, seed=100035, result=Fail
episode_id=10, seed=100036, result=Fail
episode_id=11, seed=100039, result=Success
episode_id=12, seed=100040, result=Fail
episode_id=13, seed=100041, result=Fail
episode_id=14, seed=100045, result=Fail
episode_id=15, seed=100046, result=Success
episode_id=16, seed=100049, result=Fail
episode_id=17, seed=100051, result=Fail
episode_id=18, seed=100058, result=Success
episode_id=19, seed=100063, result=Fail
episode_id=20, seed=100065, result=Success
episode_id=21, seed=100074, result=Fail
episode_id=22, seed=100075, result=Fail
episode_id=23, seed=100077, result=Fail
episode_id=24, seed=100078, result=Fail
episode_id=25, seed=100084, result=Fail
episode_id=26, seed=100089, result=Success
episode_id=27, seed=100094, result=Fail
episode_id=28, seed=100099, result=Fail
episode_id=29, seed=100101, result=Fail
episode_id=30, seed=100103, result=Fail
episode_id=31, seed=100104, result=Fail
episode_id=32, seed=100106, result=Fail
episode_id=33, seed=100107, result=Fail
episode_id=34, seed=100109, result=Success
episode_id=35, seed=100111, result=Fail
episode_id=36, seed=100113, result=Fail
episode_id=37, seed=100115, result=Success
episode_id=38, seed=100123, result=Fail
episode_id=39, seed=100124, result=Fail
episode_id=40, seed=100126, result=Fail
episode_id=41, seed=100127, result=Fail
episode_id=42, seed=100136, result=Fail
episode_id=43, seed=100141, result=Success
episode_id=44, seed=100142, result=Fail
episode_id=45, seed=100144, result=Fail
episode_id=46, seed=100146, result=Fail
episode_id=47, seed=100147, result=Fail
episode_id=48, seed=100148, result=Fail
episode_id=49, seed=100151, result=Fail
Eval log; authoritative episode records: episodes.json
episode_id=0, seed=200001, result=Fail
episode_id=1, seed=200003, result=Fail
episode_id=2, seed=200012, result=Fail
episode_id=3, seed=200016, result=Fail
episode_id=4, seed=200017, result=Success
episode_id=5, seed=200018, result=Fail
episode_id=6, seed=200022, result=Fail
episode_id=7, seed=200031, result=Fail
episode_id=8, seed=200032, result=Fail
episode_id=9, seed=200033, result=Fail
episode_id=10, seed=200038, result=Success
episode_id=11, seed=200043, result=Success
episode_id=12, seed=200044, result=Fail
episode_id=13, seed=200045, result=Fail
episode_id=14, seed=200047, result=Fail
episode_id=15, seed=200048, result=Success
episode_id=16, seed=200049, result=Fail
episode_id=17, seed=200054, result=Fail
episode_id=18, seed=200057, result=Fail
episode_id=19, seed=200058, result=Fail
episode_id=20, seed=200062, result=Fail
episode_id=21, seed=200066, result=Fail
episode_id=22, seed=200069, result=Fail
episode_id=23, seed=200072, result=Fail
episode_id=24, seed=200073, result=Fail
episode_id=25, seed=200077, result=Fail
episode_id=26, seed=200082, result=Fail
episode_id=27, seed=200083, result=Fail
episode_id=28, seed=200086, result=Fail
episode_id=29, seed=200095, result=Success
episode_id=30, seed=200100, result=Fail
episode_id=31, seed=200101, result=Fail
episode_id=32, seed=200106, result=Fail
episode_id=33, seed=200109, result=Fail
episode_id=34, seed=200110, result=Fail
episode_id=35, seed=200114, result=Fail
episode_id=36, seed=200119, result=Fail
episode_id=37, seed=200121, result=Fail
episode_id=38, seed=200122, result=Fail
episode_id=39, seed=200128, result=Fail
episode_id=40, seed=200131, result=Fail
episode_id=41, seed=200133, result=Success
episode_id=42, seed=200135, result=Fail
episode_id=43, seed=200136, result=Fail
episode_id=44, seed=200138, result=Fail
episode_id=45, seed=200142, result=Fail
End of preview. Expand in Data Studio

Mem-0 RMBench reproduction evaluation

Independent reproduction, not an official author release. 900 completed online simulation rollouts: M1 276/500 (55.2%), Mn 96/400 (24.0%). M1 uses the official retrained m1_mix executor. Mn uses official per-task executors and independently fine-tuned Qwen3-VL-8B planners.

Task Success / 100
rearrange_blocks 89
swap_blocks 77
put_back_block 100
swap_T 7
observe_and_pickup 3
cover_blocks 70
battery_try 15
blocks_ranking_try 11
press_button 0

episodes.json contains actual task/seed identities and video locations. Results include all successes and failures; replaced incomplete attempts and smoke runs are excluded. Battery evaluation resumed after a service capacity error; 38 completed episodes were retained and 62 completed subsequently. Keep task code, observation access, action budget and success criteria fixed when reusing these seeds. Historical final language strings and full initial simulator states were not saved; seeds alone do not guarantee bitwise replay.

Upstream: https://github.com/RoboTwin-Platform/RMBench (MIT). Executors: https://huggingface.co/qiuly/Mem-0-m1mix-RMBench and https://huggingface.co/qiuly/Mem-0-mn-RMBench . The executor weights are upstream dependencies, not newly trained models in this release.

Related release repositories

Downloads last month
362
Number of rows:
933
Total file size:
777 MB