> 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.datasets.llm.mock_packed

## Module Contents

### Classes

| Name                                                                                                     | Description                                                                             |
| -------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------- |
| [`MockPackedDatasetConfig`](#nemo_automodel-components-datasets-llm-mock_packed-MockPackedDatasetConfig) | Construction-time configuration for the mock packed dataset (tokenizer is a build arg). |

### Functions

| Name                                                                                               | Description                                                      |
| -------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------- |
| [`build_packed_dataset`](#nemo_automodel-components-datasets-llm-mock_packed-build_packed_dataset) | Dataset builder.                                                 |
| [`flush_block`](#nemo_automodel-components-datasets-llm-mock_packed-flush_block)                   | Flush helper (build position\_ids that reset after \<eos>).      |
| [`gen_sentence_ids`](#nemo_automodel-components-datasets-llm-mock_packed-gen_sentence_ids)         | Sentence generator with Gaussian length control.                 |
| [`make_vocab`](#nemo_automodel-components-datasets-llm-mock_packed-make_vocab)                     | Build a trivial vocab; index 0=\<pad>, 1=\<eos>, rest = word\_i. |

### Data

[`ds`](#nemo_automodel-components-datasets-llm-mock_packed-ds)

### API

```python
class nemo_automodel.components.datasets.llm.mock_packed.MockPackedDatasetConfig(
    num_blocks: int = 10,
    block_size: int = 128,
    mean_len: float = 20.0,
    std_len: float = 6.0,
    vocab_size: int = 100,
    max_sentence_len: int = 64,
    seed: int = 0
)
```

Dataclass

Construction-time configuration for the mock packed dataset (tokenizer is a build arg).

Token length of each packed block.

Maximum sentence length.

Mean sentence length (Gaussian).

Number of packed blocks to generate.

Seed for the random generator.

Standard deviation of sentence length (Gaussian).

Vocabulary size for the synthetic tokens.

```python
nemo_automodel.components.datasets.llm.mock_packed.MockPackedDatasetConfig.build(
    tokenizer: 'PreTrainedTokenizerBase | None' = None
) -> datasets.Dataset
```

Build the mock packed :class:`~datasets.Dataset` from this :class:`MockPackedDatasetConfig`.

```python
nemo_automodel.components.datasets.llm.mock_packed.build_packed_dataset(
    num_blocks: int = 10,
    block_size: int = 128,
    mean_len: float = 20.0,
    std_len: float = 6.0,
    vocab_size: int = 100,
    max_sentence_len: int = 64,
    seed: int = 0,
    tokenizer = None
)
```

Dataset builder.

```python
nemo_automodel.components.datasets.llm.mock_packed.flush_block(
    block,
    block_size
)
```

Flush helper (build position\_ids that reset after \<eos>).

```python
nemo_automodel.components.datasets.llm.mock_packed.gen_sentence_ids(
    vocab,
    mean_len: float,
    std_len: float,
    max_len: int
)
```

Sentence generator with Gaussian length control.

```python
nemo_automodel.components.datasets.llm.mock_packed.make_vocab(
    vocab_size: int = 100
)
```

Build a trivial vocab; index 0=\<pad>, 1=\<eos>, rest = word\_i.

```python
nemo_automodel.components.datasets.llm.mock_packed.ds = build_packed_dataset(num_blocks=3, block_size=32, mean_len=10, std_len=3, vocab_...
```