nemo_rl.data.hf_datasets.tulu3
#
Module Contents#
Classes#
Tulu3 preference dataset for DPO training. |
Functions#
API#
- nemo_rl.data.hf_datasets.tulu3.format_tulu3_preference(
- data: dict[str, Any],
- class nemo_rl.data.hf_datasets.tulu3.Tulu3PreferenceDataset#
Tulu3 preference dataset for DPO training.
Initialization