== test5-proxy-n8000-m8-s18
{"run": "test5-proxy-n8000-m8-s18", "paid_test_vs_start_right": [-0.0458, -0.1551, 0.0494], "questions": 24}
{"updates": 400, "approxkl_max": 0.0024, "approxkl_gt_0.05": 0, "eos_min": 60, "kl_last": 17.73}
updates 400  baseline gap -0.013  max D 0.053 at 325
ONSET none (first crossing 325, not held)
    25  judge 0.191  gold 0.204  D +0.000  tok  34.4
    50  judge 0.238  gold 0.245  D +0.006  tok  30.5
    75  judge 0.281  gold 0.266  D +0.028  tok  31.1
   100  judge 0.291  gold 0.296  D +0.009  tok  32.5
   125  judge 0.279  gold 0.276  D +0.017  tok  28.7
   150  judge 0.290  gold 0.292  D +0.010  tok  28.8
   175  judge 0.292  gold 0.286  D +0.019  tok  30.2
   200  judge 0.298  gold 0.275  D +0.036  tok  30.4
   225  judge 0.286  gold 0.288  D +0.011  tok  28.5
   250  judge 0.307  gold 0.296  D +0.024  tok  31.7
   275  judge 0.321  gold 0.304  D +0.031  tok  34.7
   300  judge 0.326  gold 0.299  D +0.040  tok  35.7
   325  judge 0.333  gold 0.293  D +0.053  tok  36.0
   350  judge 0.332  gold 0.306  D +0.040  tok  33.6
   375  judge 0.324  gold 0.315  D +0.023  tok  31.9
   400  judge 0.329  gold 0.305  D +0.037  tok  32.1
== test6-proxy-n500-m8-s18
{"run": "test6-proxy-n500-m8-s18", "paid_test_vs_start_right": [0.5805, 0.5167, 0.6364], "questions": 24}
{"updates": 400, "approxkl_max": 0.0199, "approxkl_gt_0.05": 0, "eos_min": 61, "kl_last": 42.39}
updates 400  baseline gap 0.014  max D 0.813 at 330
ONSET T = 64; 2T = 128; real run length 150
    25  judge 0.212  gold 0.199  D +0.000  tok  29.8
    50  judge 0.238  gold 0.226  D -0.002  tok  22.1
    75  judge 0.323  gold 0.234  D +0.075  tok  24.7
   100  judge 0.364  gold 0.183  D +0.168  tok  25.8
   125  judge 0.386  gold 0.112  D +0.261  tok  24.2
   150  judge 0.403  gold 0.087  D +0.302  tok  23.0
   175  judge 0.424  gold 0.075  D +0.335  tok  21.0
   200  judge 0.436  gold 0.044  D +0.378  tok  19.3
   225  judge 0.456  gold 0.039  D +0.403  tok  18.3
   250  judge 0.513  gold 0.031  D +0.468  tok  18.1
   275  judge 0.804  gold 0.001  D +0.790  tok  18.8
   300  judge 0.822  gold 0.000  D +0.808  tok  19.9
   325  judge 0.826  gold 0.000  D +0.812  tok  20.0
   350  judge 0.823  gold 0.000  D +0.809  tok  20.0
   375  judge 0.822  gold 0.000  D +0.809  tok  20.0
   400  judge 0.823  gold 0.000  D +0.809  tok  20.0
