Mistral调整了模型训练政策,现在默认将用户的输入和输出数据(包括对话和文档等)纳入模型训练计划1。这一变更对大多数用户生效,仅企业级用户在默认设置下被排除在训练之外1。
用户拥有完整的选择权来管理其数据的使用方式。在Vibe和移动应用中,用户可以通过设置面板禁用"Allow your interactions to be used to train our models"选项来退出训练计划1。Vibe和API平台的退出开关需要分别配置1。对于企业级用户,可以通过管理员面板选择加入训练计划1。
Mistral has shifted its data training policy to automatically include user inputs and outputs in model development for most customers, while excluding enterprise-tier users by default 1. Under the new arrangement, user conversations, documents, and other interaction data will be incorporated into model training unless customers actively opt out 1.
Enterprise clients are positioned differently under this policy, as they are excluded from the training program by default and must affirmatively select participation through their administrator panel if they wish their data to be used 1. For non-enterprise users seeking to disable data training, the process requires separate configuration across different platforms: the opt-out toggle is independently managed on Vibe and the mobile application through their respective settings panels, where users can disable the "Allow your interactions to be used to train our models" option 1. The same control must be configured separately for API usage 1.
评论
还没有评论,欢迎留下第一条。