Card 1.08b erratum: panel was computed before the verdict (reported separately)
Browse files
README.md
CHANGED
|
@@ -84,7 +84,7 @@ cannot be scored across the two library versions are excluded, which gives **363
|
|
| 84 |
- The p-values of the degenerate pass@1 contrasts (all zeros) were clamped to 1.0 after a boundary bug fix; no other
|
| 85 |
field changed.
|
| 86 |
|
| 87 |
-
**Standard benchmark panel (
|
| 88 |
in percent.
|
| 89 |
|
| 90 |
| | 1.07 (22327) | 1.08b | 1.08b control |
|
|
|
|
| 84 |
- The p-values of the degenerate pass@1 contrasts (all zeros) were clamped to 1.0 after a boundary bug fix; no other
|
| 85 |
field changed.
|
| 86 |
|
| 87 |
+
**Standard benchmark panel (reported separately from the verdict).** Same calibrated instrument as the 1.07 results; accuracies
|
| 88 |
in percent.
|
| 89 |
|
| 90 |
| | 1.07 (22327) | 1.08b | 1.08b control |
|