trl-lib

/

qwen1.5-1.8b-sft

like 4

Follow

TRL 36

• "torch.device",
• "transformers.integrations.deepspeed.HfTrainerDeepSpeedConfig",
• "transformers.trainer_utils.HubStrategy",
• "accelerate.utils.dataclasses.DistributedType",
• "accelerate.utils.dataclasses.DeepSpeedPlugin",
• "h4.training.configs.sft_config.SFTConfig",
• "transformers.training_args.OptimizerNames",
• "accelerate.state.PartialState",
• "transformers.integrations.deepspeed.HfDeepSpeedConfig",
• "torch.bfloat16",
• "transformers.trainer_utils.SchedulerType",
• "transformers.trainer_utils.IntervalStrategy",
• "transformers.trainer_pt_utils.AcceleratorConfig"

How to fix it?