Seat-swap comparisonspages 10

Seat-swap comparisons

10 of 10 matched pair(s), paired on instance_id, seed, arm and selected by first. 10 episode(s) left, 10 right, 10 shared key(s). The primary/USW/worst-off columns are DELTAS (right minus left), not levels.
pagemodelarmvisibilityinstancedifficultyparameter tagsseedoutcomeprimarydist NBSUSWworst-offfabricatedscore Δtotal regret
scorable_negotiation-L0-04128abf6f seed 6 · moves_only · no seat swapQwen3-8B vs Qwen3-8Bmoves_onlyFULLscorable_negotiation-L0-04128abf6f6.000deal-0.13718.0-41.00%-0.137
scorable_negotiation-L0-8069d62d1d seed 5 · moves_only · no seat swapQwen3-8B vs Qwen3-8Bmoves_onlyFULLscorable_negotiation-L0-8069d62d1d5.000deal0.302129.09.5620%0.302
scorable_negotiation-L0-9fdf5782a9 seed 2 · moves_only · no seat swapQwen3-8B vs Qwen3-8Bmoves_onlyFULLscorable_negotiation-L0-9fdf5782a92.000deal1.000365.345.30%1.000
scorable_negotiation-L0-b2f847d68b seed 2 · moves_only · no seat swapQwen3-8B vs Qwen3-8Bmoves_onlyFULLscorable_negotiation-L0-b2f847d68b2.000deal-0.126-43.063.00%-0.126
scorable_negotiation-L0-b912ac3e09 seed 3 · moves_only · no seat swapQwen3-8B vs Qwen3-8Bmoves_onlyFULLscorable_negotiation-L0-b912ac3e093.000deal0.996420.352.00%0.996
scorable_negotiation-L0-da03b2f013 seed 7 · moves_only · no seat swapQwen3-8B vs Qwen3-8Bmoves_onlyFULLscorable_negotiation-L0-da03b2f0137.000deal0.297168.056.00%0.297
scorable_negotiation-L0-e58659779a seed 3 · moves_only · no seat swapQwen3-8B vs Qwen3-8Bmoves_onlyFULLscorable_negotiation-L0-e58659779a3.000deal0.961321.338.30%0.961
scorable_negotiation-L0-e58659779a seed 4 · moves_only · no seat swapQwen3-8B vs Qwen3-8Bmoves_onlyFULLscorable_negotiation-L0-e58659779a4.000deal0.122-46.0-41.00%0.122
scorable_negotiation-L0-f91e02e13a seed 2 · moves_only · no seat swapQwen3-8B vs Qwen3-8Bmoves_onlyFULLscorable_negotiation-L0-f91e02e13a2.000deal0.975473.731.00%0.975
scorable_negotiation-L0-f91e02e13a seed 8 · moves_only · no seat swapQwen3-8B vs Qwen3-8Bmoves_onlyFULLscorable_negotiation-L0-f91e02e13a8.000deal0.975473.731.00%0.975
Click a column header to sort; / focuses the filter, Enter opens the first row that survives it.