R0: PASS — 10/10 seeds reproduce Test A-soft exactly
## Means: precision / recall / F1 / brackets chosen
3000 G: CNT 0.233/0.508/0.319/10388 HARD-R 0.239/0.165/0.195/3290 SOFT-R 0.022/0.010/0.014/2066 SOFT-RI 0.012/0.003/0.005/1368
3000 L: CNT 0.185/0.398/0.252/10257 HARD-R 0.143/0.090/0.110/2959 SOFT-R 0.011/0.003/0.005/1194 SOFT-RI 0.006/0.001/0.002/820
10000 G: CNT 0.226/0.513/0.314/10838 HARD-R 0.266/0.218/0.240/3906 SOFT-R 0.003/0.001/0.002/1923 SOFT-RI 0.001/0.000/0.000/1291
10000 L: CNT 0.183/0.412/0.253/10731 HARD-R 0.251/0.197/0.221/3750 SOFT-R 0.004/0.001/0.002/1316 SOFT-RI 0.002/0.000/0.001/892
30000 G: CNT 0.226/0.522/0.315/11030 HARD-R 0.269/0.230/0.248/4063 SOFT-R 0.000/0.000/0.000/1958 SOFT-RI 0.000/0.000/0.000/1328
30000 L: CNT 0.196/0.448/0.272/10924 HARD-R 0.235/0.189/0.209/3855 SOFT-R 0.001/0.000/0.000/1509 SOFT-RI 0.000/0.000/0.000/1029
## Predictions
G1 G: FAIL — F1 SOFT-R − CNT 3000: -0.306 [-0.311, -0.301]; 10000: -0.312 [-0.315, -0.309]; 30000: -0.315 [-0.318, -0.312]
G2 G: FAIL — F1 SOFT-R − HARD-R at 30000 -0.248 [-0.263, -0.233]
G3 G: FAIL — F1 SOFT-RI − SOFT-R at 30000 -0.000 [-0.000, +0.000]
G1 L: FAIL — F1 SOFT-R − CNT 3000: -0.248 [-0.271, -0.225]; 10000: -0.252 [-0.267, -0.237]; 30000: -0.272 [-0.281, -0.263]
G2 L: FAIL — F1 SOFT-R − HARD-R at 30000 -0.209 [-0.249, -0.168]
G3 L: FAIL — F1 SOFT-RI − SOFT-R at 30000 -0.000 [-0.000, -0.000]
G4: FAIL — F1 SOFT-RI (G) at 30000 = 0.000 (bar 0.5)