MultiOptimizer在torch1.13.1中怎么import

对于torch 1.13.1版本，我无法回答您的问题，因为该版本不存在。torch的当前最新版本是1.9.0，您可以使用以下方式导入MultiOptimizer：

from torch.optim import Adam, SGD
from transformers import AdamW, get_cosine_schedule_with_warmup, get_linear_schedule_with_warmup

class MultiOptimizer():
    def __init__(self, models, learning_rate, warmup_steps, num_training_steps):
        self.optimizers = []
        self.schedulers = []
        for model in models:
            param_optimizer = list(model.named_parameters())
            no_decay = ['bias', 'LayerNorm.bias', 'LayerNorm.weight']
            optimizer_grouped_parameters = [
                {'params': [p for n, p in param_optimizer if not any(nd in n for nd in no_decay)],
                 'weight_decay_rate': 0.01},
                {'params': [p for n, p in param_optimizer if any(nd in n for nd in no_decay)],
                 'weight_decay_rate': 0.0}
                ]
            optimizer = AdamW(optimizer_grouped_parameters, lr=learning_rate)
            self.optimizers.append(optimizer)
            self.schedulers.append(get_linear_schedule_with_warmup(
                optimizer, num_warmup_steps=warmup_steps,
                num_training_steps=num_training_steps))