← Back to context Comment by foo12bar 8 days ago Why not have a second model that reads every response looking for something nefarious? 9 comments foo12bar Reply seanhandley 7 days ago Sounds like a SIEM to me lazide 7 days ago Who watches the watcher? foo12bar 7 days ago Well you could use a model created from a completely unrelated entity. It would be difficult for an adversary to poison both in a exactly complementary way. alfiedotwtf 6 days ago But you're most likely to download both models from Huggingface lazide 7 days ago I guess it would mostly depend on training data uniqueness, yes? 4 replies →
lazide 7 days ago Who watches the watcher? foo12bar 7 days ago Well you could use a model created from a completely unrelated entity. It would be difficult for an adversary to poison both in a exactly complementary way. alfiedotwtf 6 days ago But you're most likely to download both models from Huggingface lazide 7 days ago I guess it would mostly depend on training data uniqueness, yes? 4 replies →
foo12bar 7 days ago Well you could use a model created from a completely unrelated entity. It would be difficult for an adversary to poison both in a exactly complementary way. alfiedotwtf 6 days ago But you're most likely to download both models from Huggingface lazide 7 days ago I guess it would mostly depend on training data uniqueness, yes? 4 replies →
Sounds like a SIEM to me
Who watches the watcher?
Well you could use a model created from a completely unrelated entity. It would be difficult for an adversary to poison both in a exactly complementary way.
But you're most likely to download both models from Huggingface
I guess it would mostly depend on training data uniqueness, yes?
4 replies →