LoraConfig Class
Definition
Important
Some information relates to prerelease product that may be substantially modified before it’s released. Microsoft makes no warranties, express or implied, with respect to the information provided here.
Adapter-specific metadata for LoRA models. Drives serving engine configuration at deployment time.
public class LoraConfig : System.ClientModel.Primitives.IJsonModel<Azure.AI.Projects.LoraConfig>, System.ClientModel.Primitives.IPersistableModel<Azure.AI.Projects.LoraConfig>
type LoraConfig = class
interface IJsonModel<LoraConfig>
interface IPersistableModel<LoraConfig>
Public Class LoraConfig
Implements IJsonModel(Of LoraConfig), IPersistableModel(Of LoraConfig)
- Inheritance
-
LoraConfig
- Implements
Constructors
| Name | Description |
|---|---|
| LoraConfig() |
Initializes a new instance of LoraConfig. |
Properties
| Name | Description |
|---|---|
| Alpha |
LoRA scaling factor (α). Positive integer; typically 2× the rank. |
| Dropout |
Dropout rate used during training. Informational — not used at serving time. |
| Rank |
LoRA rank (r). Positive integer. Common values: 8, 16, 32, 64. |
| TargetModules |
Model layers modified by the adapter (e.g., q_proj, v_proj). Auto-detected from adapter_config.json if omitted. |