Change Details

Per the parent task, we are just about ready to launch support for the [[ https://meta.wikimedia.org/wiki/Machine_learning_models/Production/Multilingual_revert_risk | multilingual revert risk ]] in Automoderator. We need to communicate with the wikis supported by this model that already have Automoderator deployed to let them know it is available and worth trying out. **Core info** * When the multilingual model patch is deployed, Automoderator will stop running on these wikis. * An administrator needs to go to Special:CommunityConfiguration, select Automoderator, and reconfigure and enable Automoderator * The old configuration can be viewed at [[MediaWiki:AutoModeratorConfig.json]]. * You can now opt to use the multilingual model on this wiki instead of language-agnostic. * The language-agnostic model can continue to be used, if desired. * Multilingual performance varies more by language wiki, so we couldn't implement predetermined thresholds like the wiki has used so far. * If you want to try the multilingual model, a specific threshold between 0 and 1 needs to be entered. * We have locked the threshold range between 0.85 and 1. * We recommend starting high and reducing the threshold over time until you find the right balance of true and false positives.

Per the parent task, we are just about ready to launch support for the [[ https://meta.wikimedia.org/wiki/Machine_learning_models/Production/Multilingual_revert_risk | multilingual revert risk ]] in Automoderator. We need to communicate with the wikis supported by this model that already have Automoderator deployed to let them know it is available and worth trying out. **Core info** * When the multilingual model patch is deployed, Automoderator will stop running on these wikis. * An administrator needs to go to Special:CommunityConfiguration, select Automoderator, and reconfigure and enable Automoderator * The old configuration can be viewed at [[MediaWiki:AutoModeratorConfig.json]]. * You can now opt to use the multilingual model on this wiki instead of language-agnostic. * The language-agnostic model can continue to be used, if desired. * Multilingual performance varies more by language wiki, so we couldn't implement predetermined thresholds like the wiki has used so far. * If you want to try the multilingual model, a specific threshold between 0 and 1 needs to be entered. * We have locked the threshold range between 0.85 and 1. * We recommend starting high and reducing the threshold over time until you find the right balance of true and false positives. * If you have any questions or encounter issues, please let us know.

Per the parent task, we are just about ready to launch support for the [[ https://meta.wikimedia.org/wiki/Machine_learning_models/Production/Multilingual_revert_risk | multilingual revert risk ]] in Automoderator. We need to communicate with the wikis supported by this model that already have Automoderator deployed to let them know it is available and worth trying out. **Core info** * When the multilingual model patch is deployed, Automoderator will stop running on these wikis. * An administrator needs to go to Special:CommunityConfiguration, select Automoderator, and reconfigure and enable Automoderator * The old configuration can be viewed at [[MediaWiki:AutoModeratorConfig.json]]. * You can now opt to use the multilingual model on this wiki instead of language-agnostic. * The language-agnostic model can continue to be used, if desired. * Multilingual performance varies more by language wiki, so we couldn't implement predetermined thresholds like the wiki has used so far. * If you want to try the multilingual model, a specific threshold between 0 and 1 needs to be entered. * We have locked the threshold range between 0.85 and 1. * We recommend starting high and reducing the threshold over time until you find the right balance of true and false positives. * If you have any questions or encounter issues, please let us know.