--- title: "Model Definitions" slug: "model-definitions" description: "Learn to create effective model definitions for ORCA VLMs, manage model lifecycles, and optimize document processing with our comprehensive guide." updated: 2026-10-14T19:42:58Z published: 2026-10-14T19:42:58Z canonical: "help.hyperscience.ai/model-definitions" --- > ## Documentation Index > Fetch the complete documentation index at: https://help.hyperscience.ai/llms.txt > Use this file to discover all available pages before exploring further. # Model Definitions Building an effective document-processing solution requires understanding how the components involved in model training work together. The **Model Definitions** table is designed to help manage these components. Model definitions separate configuration from the underlying models, providing a structured way to manage a model’s lifecycle. As training progresses, new model versions are created incrementally using updated training data within the same definition. > [!NOTE] > Model definitions availability > > Model definitions are currently available for VLMs, and the same framework will be used for other model types in future versions. To learn more about ORCA VLMs, see our [ORCA (Optical Reasoning and Cognition Agent) VLMs](/v44/docs/orca-vlms) article. In this article, you will learn how to: - Understand what each row in the model definitions table represents. - Create a model definition. - Understand how base-model availability and flow settings affect document processing. ## Understanding model definitions A model definition is the configuration layer that represents a specific combination of scope (what the model operates on), task (what the model does), and model type (the model architecture used). Each model definition serves as a stable reference for managing the models trained or imported for that configuration, allowing you to retrain, evaluate, and deploy new model versions over time. Each row in the Model Definitions table represents a unique combination of scope, task, and model type. A scope can include one or more objects, such as layouts, depending on the task. The remaining columns provide information about the definition’s models, including training status, deployment state, and compatibility. > [!NOTE] > Base model > > A base model provides general-purpose capabilities and is not trained on customer-specific examples. In v43.2 and later, your instance can contain multiple ORCA base models. Each specialized model remains associated with the base model used to train it, and the **History** tab shows that association. ![](https://cdn.us.document360.io/87894cef-4958-4f3f-be6f-b75a78c82548/Images/Documentation/model_definitions.png?sv=2026-02-06&spr=https&st=2026-11-09T19%3A52%3A18Z&se=2026-11-09T20%3A52%3A18Z&sr=c&sp=r&sig=JIe8nosZlff%2FgTdyp0M1V6XMwlMqat4Gi4OhrUvnhv4%3D) | Column | Description | Notes and examples | | --- | --- | --- | | **Scope** | The data or objects that the model operates on. Depending on the task, a scope can contain one or more objects, such as layouts. | For example, ORCA VLMs extract fields from documents, such as invoices. In this case, the scope of ORCA VLMs is field processing. | | **Task** | The operation that the model is trained or configured to perform within its scope. | For ORCA VLM Field Extraction, the task is field extraction. | | **Type** | The model family used for this task and scope. | For example, VLM. | | **Compatibility** | Compatibility of the most recently live model for this definition. | Compatibility depends on the model type and the base model associated with the specialized model. The supported application versions are shown in the **Compatibility** column. | | **State** | Shows whether the model is **Live** or **Inactive**. | The state is Live when the model is deployed. The state is Inactive when the model is not deployed | | **Date deployed** | The timestamp of the last deployment for this model definition. | Displays the date and time when the model was last deployed. | ## Creating a model definition This section explains how to create a model definition for ORCA VLM Field Extraction. Learn more about ORCA VLMs in our [ORCA Visual Language Models](/v44/docs/orca-vlms) article. Before you start, ensure that: - The layout you select is **Semi-structured** and contains at least one field. - The latest version of the layout is locked. - The layout is not already linked to another model definition. - Each layout can be linked to only one model definition. - A model definition can contain multiple specialized models, including ones trained with different ORCA base models. > [!NOTE] > Only one model can be live at a time > > Whether that is a specialized model or a base model. Use the interactive demo below to learn how to create a model definition: [Embedded content](https://app.storylane.io/demo/ijlnknxzdff2?embed=inline) ## Base model availability You can create an ORCA VLM Field Extraction model definition without an installed ORCA base model. However, at least one supported ORCA base model must be installed before you can train a specialized model. In v43.2 and later, multiple ORCA base models can be installed on the same instance, and the base models available to your organization may vary. Each specialized model remains associated with the ORCA base model used to train it. If a required base model is not installed: - You cannot train a specialized model using that base model. - You cannot deploy the related trained models. - You can still download or archive those model versions. - The **History** tab shows the base model as **Not installed**. Select **Install** to open **Administration > Assets**, where you can install the required base model. To install a base model, see [Installing ORCA VLMs](/v44/docs/installing-orca-vlms). > [!NOTE] > Flows and live models > > Each model definition maintains a model history containing all models trained or imported for that definition. A model definition can contain multiple specialized models, including models trained with different ORCA base models, but only **one** specialized model can be **Live** at a time. > > The **Use Live Model** setting in the ORCA extraction flow determines which model processes documents: > > - **Enabled** — The flow uses the Live specialized model in the selected model definition. The base model selected in the flow is ignored. If the model definition has no Live model, the block returns an error. > - **Disabled** — The base model selected in the flow controls processing. The flow uses a specialized model only when it was trained with the same base model. If the specialized model and selected base model do not match, the specialized model is ignored and the selected base model is used. > > Specialized models trained with ORCA 1.0 are not compatible with ORCA 2, and specialized models trained with ORCA 2 are not compatible with ORCA 1.0. > > Learn more about this setting in [Document Processing Flow](/v44/docs/document-processing-flow). ## Next steps ORCA base models provide general-purpose extraction capabilities. To adapt field extraction to your document types and business requirements: 1. Open the model definition and upload documents that represent your use case. 2. Annotate the fields you want the specialized model to extract. 3. Review the **Required Documents** indicator to determine whether the selected base model's training requirements are met. 4. Start training. - If multiple base models are installed, select the base model to use. Learn how to train a specialized model in [Training a Specialized Model](/v44/docs/training-a-specialized-model). 5. Evaluate the candidate model before deploying it as the Live model. To manage specialized models and review their base-model dependencies, see [TDM for ORCA VLMs](/v44/docs/tdm-for-orca-vlms).