课程: Responsible AI on AWS: Bedrock Guardrails, Amazon Q Security, and SageMaker Clarify

免费学习该课程！

今天就开通帐号，24,700 门业界名师课程任您挑！

Bedrock output safety: Controls and methods

Bedrock output safety: Controls and methods

课程: Responsible AI on AWS: Bedrock Guardrails, Amazon Q Security, and SageMaker Clarify

开始一个月试用 购买团队方案

Bedrock output safety: Controls and methods

“

- [Instructor] Bedrock Output Safety Controls allow for a robust security pipeline for filtering and validating a model using generated content. This infrastructure ensures that the security requirements are met before it gets into production application. So if we look at the input processing, the model is responding through the top left input channel and the input validation occurs before entering this main pipeline, and the responses are prepared for a multi-stage filtering. This is what's critical about maintaining a strict workflow, is first in the content filter module, this is going to screen for inappropriate and harmful content, and also it's configurable through strength levels, also real-time content analysis, also PII detection module. This is going to identify sensitive personal information, support blocking or masking modes, and then handle different types of PII as well. And PII stands for personally identifiable information. And then in terms of the topic filter module,…

内容