← Back to context Comment by nonethewiser 4 months ago I wonder what hooks they have in place to be able to configure safeguards at runtime. 3 comments nonethewiser Reply aleksiy123 4 months ago Probably a mix of heuristics, keywords and simple ml model.Then maybe a second gate with a lightweight llm?Edit: actually Gcp, azure, and OpenAI all have paid apis that you can also use.But I don’t think they go into details about the exact implementation https://redteams.ai/topics/defense-mitigation/guardrails-arc... ryoshu 4 months ago When we do these it's a fine-tuned classifier, generally a BERT class model. Works quite well when you sanitize input and output with low latency/cost. lyrie 4 months ago [flagged]
aleksiy123 4 months ago Probably a mix of heuristics, keywords and simple ml model.Then maybe a second gate with a lightweight llm?Edit: actually Gcp, azure, and OpenAI all have paid apis that you can also use.But I don’t think they go into details about the exact implementation https://redteams.ai/topics/defense-mitigation/guardrails-arc... ryoshu 4 months ago When we do these it's a fine-tuned classifier, generally a BERT class model. Works quite well when you sanitize input and output with low latency/cost. lyrie 4 months ago [flagged]
ryoshu 4 months ago When we do these it's a fine-tuned classifier, generally a BERT class model. Works quite well when you sanitize input and output with low latency/cost. lyrie 4 months ago [flagged]
Probably a mix of heuristics, keywords and simple ml model.
Then maybe a second gate with a lightweight llm?
Edit: actually Gcp, azure, and OpenAI all have paid apis that you can also use.
But I don’t think they go into details about the exact implementation https://redteams.ai/topics/defense-mitigation/guardrails-arc...
When we do these it's a fine-tuned classifier, generally a BERT class model. Works quite well when you sanitize input and output with low latency/cost.
[flagged]