lmflow.pipeline.utils.rm_trainer#

Classes#

`RewardTrainer`
`PeftRewardTrainer`

Functions#

`compute_metrics`(eval_pred)
`rm_loss`(model, inputs[, return_outputs])

Module Contents#

lmflow.pipeline.utils.rm_trainer.compute_metrics(eval_pred)[source]#

lmflow.pipeline.utils.rm_trainer.rm_loss(model, inputs, return_outputs=False)[source]#

class lmflow.pipeline.utils.rm_trainer.RewardTrainer[source]#

Bases: transformers.Trainer

compute_loss(model, inputs, return_outputs=False)[source]#

class lmflow.pipeline.utils.rm_trainer.PeftRewardTrainer[source]#

Bases: lmflow.pipeline.utils.peft_trainer.PeftTrainer

compute_loss(model, inputs, return_outputs=False)[source]#