books/gu-zhenren/exp15/q4a_judge.stdout.txt

42 lines
3.6 KiB
Text

########## CONTRAST floor ##########
[budget] Q4a gen spent $0.5880; judge ledger start $0.5880; global cap $2.0; contrast=floor (a0=flash_raw arm=flash_raw, seed chosen per-trap where both render)
floor a1 (s0/s1): judge diff(arm-a0) 0.2 (a0_acc 0.8 arm_acc 1.0; DET a0=fix arm=fix) spend $0.6132
floor a2 (s0/s1): judge diff(arm-a0) 0.583 (a0_acc 0.417 arm_acc 1.0; DET a0=fix arm=fix) spend $0.6346
floor b1 (s0/s1): judge diff(arm-a0) 0.375 (a0_acc 0.625 arm_acc 1.0; DET a0=fix arm=fix) spend $0.6569
floor b2 (s0/s1): judge diff(arm-a0) 0.0 (a0_acc 1.0 arm_acc 1.0; DET a0=fix arm=fix) spend $0.6821
floor c1 (s0/s1): judge diff(arm-a0) 0.075 (a0_acc 0.8 arm_acc 0.875; DET a0=fix arm=fix) spend $0.7145
floor c2 (s0/s1): judge diff(arm-a0) 0.792 (a0_acc 0.125 arm_acc 0.917; DET a0=? arm=fix) spend $0.7343
floor c3 (s0/s1): judge diff(arm-a0) -0.25 (a0_acc 1.0 arm_acc 0.75; DET a0=fix arm=fix) spend $0.7651
floor d1 (s0/s1): judge diff(arm-a0) -0.075 (a0_acc 1.0 arm_acc 0.925; DET a0=fix arm=fix) spend $0.7928
floor d2 (s0/s1): judge diff(arm-a0) 0.0 (a0_acc 1.0 arm_acc 1.0; DET a0=fix arm=fix) spend $0.8181
floor: n=9 mean diff 0.189 mean|diff| 0.261 spend $0.8181
floor: 9 scored, 0 skipped [], 0 cap-dropped []
########## CONTRAST main ##########
[budget] Q4a gen spent $0.5880; judge ledger start $0.8181; global cap $2.0; contrast=main (a0=flash_raw arm=pro_raw, seed chosen per-trap where both render)
main a1 (s2/s2): judge diff(arm-a0) 0.0 (a0_acc 1.0 arm_acc 1.0; DET a0=fix arm=fix) spend $0.8424
main a2 (s1/s1): judge diff(arm-a0) -0.958 (a0_acc 0.958 arm_acc 0.0; DET a0=fix arm=fix) spend $0.8630
main b1 (s0/s0): judge diff(arm-a0) 0.0 (a0_acc 0.75 arm_acc 0.75; DET a0=fix arm=?) spend $0.8907
main b2 (s1/s1): judge diff(arm-a0) 0.0 (a0_acc 1.0 arm_acc 1.0; DET a0=fix arm=fix) spend $0.9185
main c1 (s0/s0): judge diff(arm-a0) 0.0 (a0_acc 1.0 arm_acc 1.0; DET a0=fix arm=fix) spend $0.9517
main c2 (s0/s0): judge diff(arm-a0) 1.0 (a0_acc 0.0 arm_acc 1.0; DET a0=? arm=fix) spend $0.9705
main c3 (s0/s0): judge diff(arm-a0) 0.2 (a0_acc 0.8 arm_acc 1.0; DET a0=fix arm=fix) spend $1.0014
main d1 (s0/s0): judge diff(arm-a0) 0.0 (a0_acc 1.0 arm_acc 1.0; DET a0=fix arm=fix) spend $1.0312
main d2 (s1/s1): judge diff(arm-a0) -0.25 (a0_acc 1.0 arm_acc 0.75; DET a0=fix arm=fix) spend $1.0573
main: n=9 mean diff -0.001 mean|diff| 0.268 spend $1.0573
main: 9 scored, 0 skipped [], 0 cap-dropped []
########## CONTRAST editor ##########
[budget] Q4a gen spent $0.5880; judge ledger start $1.0573; global cap $2.0; contrast=editor (a0=pro_raw arm=pro_glm, seed chosen per-trap where both render)
editor a1 (s2/s2): judge diff(arm-a0) 0.0 (a0_acc 1.0 arm_acc 1.0; DET a0=fix arm=?) spend $1.0825
editor a2 (s1/s1): judge diff(arm-a0) -0.05 (a0_acc 0.25 arm_acc 0.2; DET a0=fix arm=fix) spend $1.1151
editor b1 (s0/s0): judge diff(arm-a0) 0.0 (a0_acc 0.75 arm_acc 0.75; DET a0=? arm=?) spend $1.1436
editor b2 (s1/s1): judge diff(arm-a0) 0.0 (a0_acc 1.0 arm_acc 1.0; DET a0=fix arm=fix) spend $1.1668
editor c1 (s0/s0): judge diff(arm-a0) 0.0 (a0_acc 1.0 arm_acc 1.0; DET a0=fix arm=fix) spend $1.1971
editor c2 (s0/s0): judge diff(arm-a0) 0.0 (a0_acc 1.0 arm_acc 1.0; DET a0=fix arm=fix) spend $1.2261
editor c3 (s0/s0): judge diff(arm-a0) 0.0 (a0_acc 1.0 arm_acc 1.0; DET a0=fix arm=fix) spend $1.2570
editor d1 (s0/s0): judge diff(arm-a0) 0.05 (a0_acc 0.95 arm_acc 1.0; DET a0=fix arm=fix) spend $1.2861
editor d2 (s1/s1): judge diff(arm-a0) -0.917 (a0_acc 1.0 arm_acc 0.083; DET a0=fix arm=fail) spend $1.3036
editor: n=9 mean diff -0.102 mean|diff| 0.113 spend $1.3036
editor: 9 scored, 0 skipped [], 0 cap-dropped []