In the Azure portal, go to your Personalizer resource's Configuration page, and change the Reward wait time to 10 minutes. This short duration will train the model rapidly, allowing you to see how the recommended action changes for each iteration. In the Azure portal, go to your Personalizer resource's Configuration page, and change the Model update frequency to 30 seconds. This will update your Personalizer instance to be a 'Multi Slot' Personalizer and will now support multi-slot Rank and Reward calls. In the same tab in the portal, under import learning settings browse to find your recently modified json file and upload it. ADF (action dependent features) means that the actions are expressed / identified with features. CB (contextual bandits) and CCB (conditional contextual bandits) are the algorithms Personalizer uses for single-slot and multi-slot personalization, respectively. Change this to -ccb_explore_adf and save the file. The arguments field in the downloaded json file will start with -cb_explore_adf. In the Azure portal, in the Personalizer resource, under Resource Management, on the Model and learning settings page, select Export learning settings. Automatic Optimization for multi-slot personalization will be supported in the future. Multi-slot personalization will not work unless you disable Automatic Optimization.
