tokenizer = BertTokenizer.from_pretrained('RoBERTa/vocab.txt')
config = BertConfig.from_pretrained('RoBERTa/config.json')
model = BertForSequenceClassification.from_pretrained('RoBERTa/pytorch_model.bin', config=config)
pytorch用BERT的加载方式加载roberta模型,呢么创建token时special token 是按照bert的方式还是roberta的方式呢
pytorch是否有快速加载模型的方法呢?
类似:
tokenizer = BertTokenizer(pretrained_roberta_name)
bert = BertModel.from_pretrained(pretrained_roberta_name)
tokenizer = BertTokenizer.from_pretrained('RoBERTa/vocab.txt')
config = BertConfig.from_pretrained('RoBERTa/config.json')
model = BertForSequenceClassification.from_pretrained('RoBERTa/pytorch_model.bin', config=config)
pytorch用BERT的加载方式加载roberta模型,呢么创建token时special token 是按照bert的方式还是roberta的方式呢
pytorch是否有快速加载模型的方法呢?
类似:
tokenizer = BertTokenizer(pretrained_roberta_name)
bert = BertModel.from_pretrained(pretrained_roberta_name)