nemo_automodel.components.datasets.llm.mock

Module Contents

Classes

Name	Description
`MockUnpackedDatasetConfig`	Construction-time configuration for the mock unpacked dataset (tokenizer is a build arg).

Functions

Name	Description
`build_unpacked_dataset`	Build a dataset where each example is one sentence (variable length).
`gen_sentence_ids`	Sentence generator with Gaussian length control.
`make_vocab`	Build a trivial vocab; index 0=<pad>, 1=<eos>, rest = tok_i.

Data

ds

API

class nemo_automodel.components.datasets.llm.mock.MockUnpackedDatasetConfig(
    num_sentences: int = 10,
    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 unpacked dataset (tokenizer is a build arg).

accepts_tokenizer

bool = True

max_sentence_len

int = 64

Maximum sentence length.

mean_len

float = 20.0

Mean sentence length (Gaussian).

num_sentences

int = 10

Number of sentence examples to generate.

seed

int = 0

Seed for the random generator.

std_len

float = 6.0

Standard deviation of sentence length (Gaussian).

vocab_size

int = 100

Vocabulary size for the synthetic tokens.

nemo_automodel.components.datasets.llm.mock.MockUnpackedDatasetConfig.build(
    tokenizer: 'PreTrainedTokenizerBase | None' = None
) -> datasets.Dataset

Build the mock unpacked :class:~datasets.Dataset from this :class:MockUnpackedDatasetConfig.

nemo_automodel.components.datasets.llm.mock.build_unpacked_dataset(
    num_sentences: int = 10,
    mean_len: float = 20.0,
    std_len: float = 6.0,
    vocab_size: int = 100,
    max_sentence_len: int = 64,
    seed: int = 0,
    tokenizer = None
)

Build a dataset where each example is one sentence (variable length).

Returns:

a HuggingFace Dataset with fields: input_ids: Sequence(int64) attention_mask:Sequence(int8) labels: Sequence(int64) position_ids: Sequence(int64)

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

Sentence generator with Gaussian length control.

nemo_automodel.components.datasets.llm.mock.make_vocab(
    vocab_size: int = 100
)

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

nemo_automodel.components.datasets.llm.mock.ds = build_unpacked_dataset(num_sentences=5, mean_len=12.0, std_len=3.0, vocab_size=5...