lmflow.pipeline.iterative_dpo_aligner#

Attributes#

logger

Classes#

IterativeDPOAligner

Module Contents#

lmflow.pipeline.iterative_dpo_aligner.logger[source]#
class lmflow.pipeline.iterative_dpo_aligner.IterativeDPOAligner(model_args: lmflow.args.ModelArguments, data_args: lmflow.args.DatasetArguments, aligner_args: lmflow.args.IterativeDPOAlignerArguments, ref_model_args: lmflow.args.ModelArguments, reward_model_args: lmflow.args.ModelArguments, **kwargs)[source]#
model_args[source]#
data_args[source]#
aligner_args[source]#
ref_model_args[source]#
reward_model_args[source]#
workspace_path[source]#
align(dataset_list: list[lmflow.datasets.dataset.Dataset])[source]#