# Think Paragraph Configuration You can set think paragraph configuration by clicking on the **'Configuration'** ()on the top right of the page. Then, you can see a pop-up of configuration and you can decide how exactly your bot interprets what your customer says by using confidence level. Every time your customer types a message, the bot analyses its accuracy. In this pop-up, you can also set the client id and client secret. ![Think Paragraph Configuration](https://765826444-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FLYDAnLjpg2T8z2BplJOS%2Fuploads%2FpDtmognqNA9VMQC6yPjf%2FThink%20Paragrap%20Configuration%20edit.png?alt=media\&token=27cebb21-e76a-4c99-b27b-c4e02fd17db0) **Component Explanation:** | Name | Description | | ---------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Retrieval Score (Confidence Level) | You can decide how precisely your bot interprets what your customer says by using confidence level. Every time your customer types a message, the bot analysis its accuracy. | | Client Id & Client Secret | Client id and client secret used for integration with 3Dolphins Think. | | Iteration | To determine the number of epoch training to be run. | | Learning Rate | The amount that the weight is updated during the training. | | Min Learning Rate | Setting the minimum amount that the weight is updated during the training. | | Layer Size | Parameter to configure the size of hidden layer. | | Batch Size | Specifies the number of samples to process before updating the internal model parameters (the number of words processed at once). | | Min Word Frequency | The minimum number of times a word must appear in the corpus. | | Window Size | To set the length of a cutout of a time sequence of data. |