Setting The Fire¶
Agent runs
Each mode of this task runs once per run. Est. cost is tokens at list price (data/prices.yml), never a bill; see the BEHAVIOR-1K runs for every task of a run.
Codex + GPT-6 Luna, reasoning xhigh (ChatGPT login)default run
| Trial | Started | Agent time | Model requests | Tokens in / out | Est. cost | Grade |
|---|---|---|---|---|---|---|
| U✗ 4h 00m 0% | 09-27 08:03 | 4h 00m | 731 | 89.3M / 418k | $1.24 | grader_error 0, missing_trajectory 1 |
| L✗ 11m 0% | 09-27 10:41 | 11m | 26 | 391k / 4k | $0.0089 | deterministic 0, grader_error 0, live_success 0, n_actions 0, replay_success 0, video_rendered 0 |
Task instruction (upstream)
In the living room, place the newspaper from the table into the wood fireplace, then put both pieces of firewood from the floor on top of the newspaper. Use the cigar lighter to ignite any one of the items so that the whole pile catches fire, and then turn the lighter off.
Measured on the stack we run — BEHAVIOR-1K v3.9.2 / OmniGibson 3.9.2 / Isaac Sim 5.1
| Goal predicates | inside on_fire ontop toggled_on |
| Quantifiers | forall |
| Goal clauses | 3 |
| Objects named in the goal | cigar_lighter.n.01 firewood.n.01 newspaper.n.03 wood_fireplace.n.01 |
| Objects in the problem | 8 in 7 categories |
| Rooms loaded | corridor_0 garden_0 kitchen_0 living_room_0 |
| Demonstrations | 200 teleoperated episodes |
| Mean episode | 5m 04s (9,118 control steps at 30 Hz) |
| Base travel, mean | 31.8 m |
| Gripper travel, mean | left 11.4 m · right 12.95 m |
| Evaluation instances | 20 public test instances (ids 301–320) |
Read from BDDL activity definitions + 2026-challenge-task-instances (licensed download), 2026-09-21.
Tags¶
Task DomainMobile / Whole-body Manipulation
Why this task is interesting¶
Not yet written.
Capability notes¶
Not yet written.
Oracle demo review¶
Not yet reviewed.