FAQ

• 1、分布式训练报错This error indicates that your module has parameters that were not used in producing loss
  A:1.使用下面代码打印所有需要计算梯度的参数，确认没有问题 2.如果有问题，修改问题，如果没有问题，passing the keyword argument find_unused_parameters=True to torch.nn.parallel.DistributedDataParallel

    enabled = {}
    for name, param in model_without_ddp.named_parameters():
        if param.requires_grad:
            enabled[name] = param.shape
    print(f"Parameters to be updated: {enabled.keys()}")