Skip to content

Under maintenance. Some results may be incomplete or inaccurate.

Robotensor

Vector: one demonstration from another scene, level 1

Duel

Crown taken23 Sept 2026, 13:44 UTClight duel0c7c1d2925…a49b48

Reigning model

robotensor/vector-base

Revision
741365e9d895f84cbc67f4982189d77ec8b3d822
Submission key
43c303d383266fbd

12.8%

Average over the skills

Sixteen tabletop tasks (two Franka arms, level 1)

Challenger

Holds the crown

robotensor/vector-update1

Revision
c9e5372137ac9d226d4a31a9319a1c670925d1d3
Submission key
a2f91f76022456c2

38.3%

Average over the skills

Sixteen tabletop tasks (two Franka arms, level 1)
Margin check38.3% − 12.8% = +25.5 ptsneeds ≥ +3.0 ptsMet

Verdict derivation

recomputed from 48 unitsCrown taken

The ledger

Per-skill success rates for both sides and their difference
SkillReigningChallengerΔ
Sixteen tabletop tasks (two Franka arms, level 1)12.8%6 / 4738.3%18 / 47+25.5 pts
Averageneeds ≥ 15.8%12.8%38.3%+25.5 pts

challenger average ≥ reigning average + +3.0 pts

38.3% ≥ 12.8% + +3.0 pts · needs ≥ 15.8%

Met

The crown moved to the challenger

38.3% vs 12.8% · +25.5 pts

Paired outcomes

diagnostic, does not decide the crown

Won

12

Lost

0

Tied

35

Decided

12

Void

1

The paired record

48 units · 12 decided · 35 tied · 1 void

  • Reigning
    12.8%6/47
    Challenger
    38.3%18/47
    Δ
    +25.5 pts
  • Reigning
    0.0%0/3
    Challenger
    66.7%2/3
  • Tied
    • scene 674703748
    • demo seeds
    • embodiment franka-panda
    • level 1
    • scene seeds
    • task config level1
    Reigning
    Fail
    Challenger
    Fail
  • Challenger
    • scene 53353137
    • demo seeds
    • embodiment franka-panda
    • level 1
    • scene seeds
    • task config level1
    Reigning
    Fail
    Challenger
    Pass
  • Challenger
    • scene 659278052
    • demo seeds
    • embodiment franka-panda
    • level 1
    • scene seeds
    • task config level1
    Reigning
    Fail
    Challenger
    Pass
  • Reigning
    0.0%0/3
    Challenger
    0.0%0/3
  • Reigning
    0.0%0/3
    Challenger
    0.0%0/3
  • Reigning
    0.0%0/3
    Challenger
    66.7%2/3
  • Reigning
    0.0%0/3
    Challenger
    66.7%2/3
  • Reigning
    0.0%0/3
    Challenger
    0.0%0/3
  • Reigning
    66.7%2/3
    Challenger
    100.0%3/3
  • Reigning
    0.0%0/3
    Challenger
    0.0%0/3
  • Reigning
    66.7%2/3
    Challenger
    100.0%3/3
  • Reigning
    0.0%0/3
    Challenger
    33.3%1/3
  • Reigning
    0.0%0/3
    Challenger
    33.3%1/3
  • Reigning
    0.0%0/3
    Challenger
    33.3%1/3
  • Reigning
    33.3%1/3
    Challenger
    33.3%1/3
  • Reigning
    33.3%1/3
    Challenger
    66.7%2/3
  • Reigning
    0.0%0/3
    Challenger
    0.0%0/3
  • Reigning
    0.0%0/2
    Challenger
    0.0%0/2
  • Crown rule: margin-met (+25.53 points against a margin of 3; sign test p=0.0002441 against alpha 0.05).

Provenance

Record
0c7c1d29257b…4ea49b48
Kind
Duel
Finished
23 Sept 2026, 13:44 UTC
Wall time
2h 16m
Duel size
light
Units per skill
48 × 1 skills = 48 per side
Score margin in force
+3.0 pts
Duel id
f958a150021b…344e9332
Protocol version
8
Protocol fingerprint
484ed1fa124f…ae09a6f0
Event digest, as signed
83b821d11435…9874b6cc
Cycle
2
Media attached
143 clips