`nemo_rl.data.hf_datasets.deepscaler`#

Module Contents#

`format_math`
`prepare_deepscaler_dataset`	Load and split the DeepScaler dataset into train and test sets.

nemo_rl.data.hf_datasets.deepscaler.format_math( data: dict[str, str | float | int], ) → dict[str, list[Any] | str][source]#

nemo_rl.data.hf_datasets.deepscaler.prepare_deepscaler_dataset( seed: int = 42, ) → dict[str, datasets.Dataset | None][source]#: Load and split the DeepScaler dataset into train and test sets.

class nemo_rl.data.hf_datasets.deepscaler.DeepScalerDataset(seed: int = 42)[source]#

Initialization

Initialize the DeepScaler dataset with train/test split.