As stated, it's work product. Not a toy "adversarial prompt," so I won't be posting it here.
And as stated, even Pangram's own marketing material claims at least a 0.4% false negative rate on their own contrived corpus. A one in 250 event is hardly even rare.
You seem quite defensive about Pangram, and it shows up in a few of your comments. Is there a reason for that? If it's extending to outright denial of Pangram's own stated error rates... sheesh... very weird indeed (unless they're paying you, obviously).
And as stated, even Pangram's own marketing material claims at least a 0.4% false negative rate on their own contrived corpus. A one in 250 event is hardly even rare.
You seem quite defensive about Pangram, and it shows up in a few of your comments. Is there a reason for that? If it's extending to outright denial of Pangram's own stated error rates... sheesh... very weird indeed (unless they're paying you, obviously).