> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/nemo/automodel/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/nemo/automodel/_mcp/server.

# nemo_automodel.components.models.inkling.processing

Processor construction helpers for Inkling fine-tuning.

## Module Contents

### Functions

| Name                                                                                                      | Description                                                            |
| --------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------- |
| [`build_inkling_processor`](#nemo_automodel-components-models-inkling-processing-build_inkling_processor) | Load Inkling's processor and configure padding with an existing token. |

### Data

[`_INKLING_END_OF_SAMPLING_TOKEN`](#nemo_automodel-components-models-inkling-processing-_INKLING_END_OF_SAMPLING_TOKEN)

### API

```python
nemo_automodel.components.models.inkling.processing.build_inkling_processor(
    pretrained_model_name_or_path: str,
    kwargs: typing.Any = {}
) -> transformers.processing_utils.ProcessorMixin
```

Load Inkling's processor and configure padding with an existing token.

The published tokenizer does not declare EOS or padding tokens even though
its chat template terminates assistant responses with an existing
end-of-sampling token. Reusing that token keeps the checkpoint vocabulary
unchanged and enables padded fine-tuning batches.

**Parameters:**

Hugging Face model ID or local snapshot.

Additional arguments forwarded to `AutoProcessor.from_pretrained`.

**Returns:** `ProcessorMixin`

The configured Inkling processor.

```python
nemo_automodel.components.models.inkling.processing._INKLING_END_OF_SAMPLING_TOKEN = '<|content_model_end_sampling|>'
```