python 3.12.13, sentence-transformers 6.2.0, transformers 5.22.1, torch 2.15.0, numpy 2.5.4, Darwin arm64
corpus paragraphs: 1439
{'val': 1000, 'test': 1000}
right one in top 3   n=414  top-1 score median 0.605
right one missed     n= 86  top-1 score median 0.561
no answer in corpus  n=500  top-1 score median 0.349
t=0.3  precision 0.465  coverage 0.886
t=0.4  precision 0.660  coverage 0.608
t=0.5  precision 0.802  coverage 0.434
t=0.6  precision 0.863  coverage 0.255
t=0.7  precision 0.921  coverage 0.063
chosen on validation: t=0.49  ->  test precision 0.676, coverage 0.490
score 0.2-0.3: n=114  right paragraph in top 3: 2%
score 0.3-0.4: n=278  right paragraph in top 3: 4%
score 0.4-0.5: n=174  right paragraph in top 3: 30%
score 0.5-0.6: n=179  right paragraph in top 3: 72%
score 0.6-0.7: n=192  right paragraph in top 3: 84%
score 0.7-0.8: n= 59  right paragraph in top 3: 93%
bge median top score, no answer in corpus: 0.584
bge with MiniLM's cutoff 0.49: test precision 0.410, coverage 0.990
bge with its own cutoff 0.67: test precision 0.740, coverage 0.470
t=0.49 with ~10% unanswerable: precision 0.792, coverage 0.752

MiniLM, 80% target, other random article splits (seed 0 is the worked example above):
split seed 0: cutoff 0.49  validation precision 0.801  test precision 0.676  test coverage 0.490
split seed 1: cutoff 0.48  validation precision 0.812  test precision 0.754  test coverage 0.464
split seed 2: cutoff 0.58  validation precision 0.804  test precision 0.844  test coverage 0.307
split seed 3: cutoff 0.55  validation precision 0.802  test precision 0.882  test coverage 0.331
split seed 4: cutoff 0.48  validation precision 0.805  test precision 0.727  test coverage 0.473
split seed 5: cutoff 0.50  validation precision 0.803  test precision 0.779  test coverage 0.425
split seed 6: cutoff 0.53  validation precision 0.800  test precision 0.825  test coverage 0.382
split seed 7: cutoff 0.52  validation precision 0.806  test precision 0.830  test coverage 0.412
split seed 8: cutoff 0.56  validation precision 0.801  test precision 0.821  test coverage 0.352
split seed 9: cutoff 0.53  validation precision 0.801  test precision 0.876  test coverage 0.395
test precision over 10 splits: min 0.676, median 0.823, max 0.882
total run time: 122.4 s