`nemo_rl.data.datasets.response_datasets.oasst`#

Module Contents#

Simple wrapper around the OASST dataset.

`parse_conversations`	Recusive function that returns all the sub converstaions in a list starting from node tree_obj.
`get_data_records`

nemo_rl.data.datasets.response_datasets.oasst.SYSTEM_PROMPT = <Multiline-String>#

nemo_rl.data.datasets.response_datasets.oasst.parse_conversations(tree_obj, first: bool = False)#

Recusive function that returns all the sub converstaions in a list starting from node tree_obj.

Parameters:: tree_obj (obj) – current conversation node
Returns:: a list of sub conversation threads including the current conversation node

nemo_rl.data.datasets.response_datasets.oasst.get_data_records(objs, task_name: str = 'oasst')#

class nemo_rl.data.datasets.response_datasets.oasst.OasstDataset(

split_validation_size: float = 0.05,

seed: int = 42,

**kwargs,

)#

Simple wrapper around the OASST dataset.

Parameters:

split_validation_size – Size of the validation data, default is 0.05
seed – Seed for train/validation split when split_validation_size > 0, default is 42

Initialization