core.trainers.trl
core.trainers.trl
Module for TRL PPO trainer
Classes
Name | Description |
---|---|
AxolotlCPOTrainer | Extend the base CPOTrainer for axolotl helpers |
AxolotlKTOTrainer | Extend the base KTOTrainer for axolotl helpers |
AxolotlORPOTrainer | Extend the base ORPOTrainer for axolotl helpers |
AxolotlPRMTrainer | Extend the base trl.PRMTrainer for axolotl helpers |
AxolotlRewardTrainer | Extend the base RewardTrainer for axolotl helpers |
TRLPPOTrainer | Wrapper for TRL PPO trainer to handle customizations |
AxolotlCPOTrainer
*args, **kwargs) core.trainers.trl.AxolotlCPOTrainer(
Extend the base CPOTrainer for axolotl helpers
AxolotlKTOTrainer
*args, **kwargs) core.trainers.trl.AxolotlKTOTrainer(
Extend the base KTOTrainer for axolotl helpers
AxolotlORPOTrainer
*args, **kwargs) core.trainers.trl.AxolotlORPOTrainer(
Extend the base ORPOTrainer for axolotl helpers
AxolotlPRMTrainer
*args, **kwargs) core.trainers.trl.AxolotlPRMTrainer(
Extend the base trl.PRMTrainer for axolotl helpers
AxolotlRewardTrainer
*args, **kwargs) core.trainers.trl.AxolotlRewardTrainer(
Extend the base RewardTrainer for axolotl helpers
TRLPPOTrainer
core.trainers.trl.TRLPPOTrainer()
Wrapper for TRL PPO trainer to handle customizations