abideen

/

Mistral-v2-orpo

like 1

• "torch.BFloat16Storage",
• "torch._tensor._rebuild_from_type_v2",
• "collections.OrderedDict",
• "torch._utils._rebuild_tensor_v2",
• "torch.Tensor"

How to fix it?

• "torch.device",
• "accelerate.utils.dataclasses.DistributedType",
• "transformers.trainer_utils.HubStrategy",
• "transformers.trainer_utils.IntervalStrategy",
• "transformers.trainer_utils.SchedulerType",
• "transformers.training_args.TrainingArguments",
• "accelerate.state.PartialState",
• "transformers.trainer_pt_utils.AcceleratorConfig",
• "transformers.training_args.OptimizerNames"

How to fix it?