(Optional) Creating and Managing Models on KooSearch

Scenarios

You can configure the models you need on the model management page. When you create a knowledge base, you can select the models you want to use with it. Or you can choose models when you trial-use KooSearch's intelligent Q&A and search services to get better results.

Creating a Model

1. Assess the KooSearch console.
2. In the navigation pane on the left, choose Configuration Management > Model Management. The Model Management page is displayed.
3. Click Create Model.
   Figure 1 Creating a model
   

4. On the displayed page, set the parameters described in the following table. Then click OK.

   Table 1 Creating a model

   Parameter	Description
   Model Name	Enter a model name. The value cannot be empty.
   Model Type	NLP models - Cloud base: Pangu NLP models provided by Huawei Cloud. NLP models - Bare metal: Pangu NLP models deployed on bare metal servers. Search embedding model: a vector search model that converts text into numerical representations called embeddings, which are essentially vectors. Search reranking model: Reranks the search results to provide more relevant results. Search planning model: Provides multi-turn rewriting and intent recognition. Moderation model: Provides a content moderation service that checks the compliance of questions and answers. Only one moderation model can be created. OCR model: Extracts text from images, scanned documents, PDFs, and OFD files and outputs the text in an editable format. NLP Model - Ascend Cloud: The NLP model is available as a MaaS service on the Ascend cloud. If you select this model to provide Q&A, you are advised to set the maximum number of new tokens generated by the model in response to a given input to 512. Cache generation model: Calculates the similarity between queries and provides a cache for the knowledge base. Web search engine service: Allows users to add a custom search engine service. Enhanced web search service: Provides an enhanced web search service. NOTE: There is a strong connection between the embedding model and the cache generation model. When an embedding model is created, the system automatically generates a cache generation model. If any configuration information is deleted by mistake, the model must be rebuilt using the same configuration parameters. For example, if the name of the embedding model is pangu_embedding, the name of the matching cache generation model is pangu_embedding_faq. When creating a knowledge base, both the embedding model (pangu_embedding) and cache generation model (pangu_embedding_faq) are required. If the cache generation model (pangu_embedding_faq) does not exist or is not accessible, an error is returned. In this case, the administrator needs to check whether the pangu_embedding_faq model exists or whether the knowledge base user has access to it. If the model is missing, create it. If the knowledge base user does not have access to it, grant them the access permission.
   Access Address	Internal network address and port number of the model.
   Enable	If you select Review model, the Enable button is displayed. When Enable is selected, the Moderation Model will check the compliance of questions asked by users on the Experience Platform as well as the answers generated by AI. KooSearch will refuse to answer questions that contain sensitive words by returning a preset response.
   Description	A detailed description of the model.
   Ascend Cloud Model Name	If Model Type is set to NLP Model-Ascend Cloud, you need to specify Ascend Cloud Model Name. Select an NLP model provided via the Ascend AI Cloud Service.
   Context Length (K)	If Model Type is set to NLP model-Cloud base or NLP model-Bare Metal, you need to set Context Length. Context length refers to the maximum number of tokens the NLP model can process in a single input sequence. A larger context window enables an AI model to process longer inputs and generate more comprehensive outputs.
   Deployment ID	If Model Type is set to NLP model-Cloud base or NLP model-Bare Metal, you need to set Deployment ID. Deployment ID indicates the model deployment ID.
   Authentication Type	IAM authentication: Huawei IAM authentication is supported. By default, the CSS resource tenant is authenticated. When entrusted account authentication is enabled, you may use an entrusted account to perform the authentication.
   	Custom authentication: Custom request headers can be added during API calling.
   URL	API management address of the embedding or reranking model when you configure the KooSearch dependent cluster. You can obtain the URL from the dedicated cluster.
   Enable Periodic Detection	Periodically checks connectivity to all models and updates the status to the model list.

5. Click OK. If you have selected an NLP model, the Statement dialog box shown below is displayed. Select the check box below to agree to the statement, and click Confirm.
   Figure 2 Disclaimer
   

6. After the model is created, find it on the model management page. You can click the model name to check its information.

Editing and Deleting a Model

1. Assess the KooSearch console.
2. In the navigation pane on the left, choose Model Management. The Model Management page is displayed.
3. Select the target model.

   Click Edit in the Operation column to edit the model. For details about its parameters, see Table 1.

   Click Delete in the Operation column to delete a model.