19 Commits

Author	SHA1	Message	Date
Brandon Thomas	e43580d627	tests	2024-03-03 22:43:10 -05:00
Brandon Thomas	4c1b8f1d4c	refactor prompt classification	2024-03-03 22:20:36 -05:00
Brandon Thomas	6c6debe64b	additional safety features	2024-02-24 00:35:26 -05:00
Brandon Thomas	c6555b88d5	one or two more terms	2024-02-21 05:14:32 -05:00
Brandon Thomas	2dab930226	additional test cases for italian	2024-02-20 21:40:08 -05:00
Brandon Thomas	9fc6e4adf6	split up tests	2024-02-20 21:28:23 -05:00
Brandon Thomas	4ae7e61b7e	additional safety	2024-02-20 19:26:21 -05:00
Brandon Thomas	16a3f74105	"1girl" is a false positive and used frequently in safe contexts	2024-02-20 19:09:03 -05:00
Brandon Thomas	4ad90b31df	stop racism too	2024-02-20 18:56:24 -05:00
Brandon Thomas	5cc90d1001	better classifier	2024-02-20 18:49:20 -05:00
Brandon Thomas	688efffcb8	improve classifier	2024-02-20 18:37:01 -05:00
Brandon Thomas	57956c8124	fix failing lints and tests	2024-02-20 02:44:43 -05:00
Brandon Thomas	ea269459e6	more prompt tests	2024-02-20 01:49:03 -05:00
Brandon Thomas	716589f62c	more safety + tests	2024-02-19 19:55:06 -05:00
Brandon Thomas	2d5188ba91	more dictionary terms and test cases	2024-02-19 19:37:03 -05:00
Brandon Thomas	d5205e764e	more prompt tests, better regex split	2024-02-19 18:54:13 -05:00
Brandon Thomas	c87b1ba20d	filter more bad prompts	2024-02-19 12:55:01 -05:00
Brandon Thomas	8dc3d29daf	Block basic absuive SD prompts	2024-02-19 02:31:07 -05:00
Brandon Thomas	bd9c4d752f	move includes to single root dir	2023-08-15 23:42:01 -04:00