|
| 1 | +--- |
| 2 | +subcategory: "inference" |
| 3 | +page_title: "Scaleway: scaleway_inference_model" |
| 4 | +--- |
| 5 | + |
| 6 | +# scaleway_inference_model |
| 7 | + |
| 8 | +The `scaleway_inference_model` data source allows you to retrieve information about an inference model available in the Scaleway Inference API, either by providing the model's `name` or its `model_id`. |
| 9 | + |
| 10 | +## Example Usage |
| 11 | + |
| 12 | +### Basic |
| 13 | + |
| 14 | +```hcl |
| 15 | +data "scaleway_inference_model" "my_model" { |
| 16 | + name = "meta/llama-3.1-8b-instruct:fp8" |
| 17 | +} |
| 18 | +``` |
| 19 | + |
| 20 | +## Argument Reference |
| 21 | + |
| 22 | +You must provide either name or model_id, but not both. |
| 23 | + |
| 24 | +- `name` (Optional, Conflicts with model_id) The fully qualified name of the model to look up (e.g., "meta/llama-3.1-8b-instruct:fp8"). The provider will search for a model with an exact name match in the selected region and project. |
| 25 | +- `model_id` (Optional, Conflicts with name) The ID of the model to retrieve. Must be a valid UUID with locality (i.e., Scaleway's zoned UUID format). |
| 26 | +- `project_id` (Optional) The project ID to use when listing models. If not provided, the provider default project is used. |
| 27 | +- `region` (Optional) The region where the model is hosted. If not set, the provider default region is used. |
| 28 | + |
| 29 | +## Attributes Reference |
| 30 | + |
| 31 | +In addition to the input arguments above, the following attributes are exported: |
| 32 | + |
| 33 | +- `id` - The unique identifier of the model. |
| 34 | +- `tags` - Tags associated with the model. |
| 35 | +- `status` - The current status of the model (e.g., ready, error, etc.). |
| 36 | +- `description` - A textual description of the model (if available). |
| 37 | +- `has_eula` - Whether the model requires end-user license agreement acceptance before use. |
| 38 | +- `parameter_size_bits` - Size, in bits, of the model parameters. |
| 39 | +- `size_bytes` - Total size, in bytes, of the model archive. |
| 40 | +- `nodes_support` - List of supported node types and their quantization options. Each entry contains: |
| 41 | + - `node_type_name` - The type of node supported. |
| 42 | + - `quantization` - A list of supported quantization options, including: |
| 43 | + - `quantization_bits` - Number of bits used for quantization (e.g., 8, 16). |
| 44 | + - `allowed` - Whether this quantization is allowed. |
| 45 | + - `max_context_size` - Maximum context length supported by this quantization. |
| 46 | + |
| 47 | + |
0 commit comments