使用Transformers微调机器翻译预训练模型
我的数据为两列["zh","en"]
在分词的时候,使用的语句是
def tokenize_function(examples):
return tokenizer(examples["zh"],examples["en"] padding="max_length", truncation=True)
分词效果如下
已经存在input_ids
为什么最后调用traner.train()函数会报如下错误:
You have to specify either decoder_input_ids or decoder_inputs_embeds
麻烦大家帮帮忙