3 ms·
We're using small language models to detect prompt injection. Not too cool, but at least we can publish some AI-related stuff on the internet without a huge bil
by eb0la 2y ago
We're using small language models to detect prompt injection. Not too cool, but at least we can publish some AI-related stuff on the internet without a huge bill.
- sitkack 2y agoWhat kind of prompt injection attacks do you filter out? Have you tested with a prompt tuning framework?