dbiir / UER-py

Open Source Pre-training Model Framework in PyTorch & Pre-trained Model Zoo
https://github.com/dbiir/UER-py/wiki
Apache License 2.0
2.97k stars 528 forks source link

利用官方BERT进行下游任务(多分类)微调时报错 #298

Closed Moximixi closed 2 years ago

Moximixi commented 2 years ago

想要利用bert模型进行六分类下游任务微调,但是报错 运行命令:

CUDA_VISIBLE_DEVICES=0  python3 finetune/run_classifier.py --pretrained_model_path models/bert-base-chinese/pytorch_model.bin \
                                   --vocab_path models/bert-base-chinese/bert-base-chinese-vocab.txt \
                                   --train_path datasets/weibo/train.tsv \
                                   --dev_path datasets/weibo/dev.tsv \
                                   --test_path datasets/weibo/test.tsv \
                                   --output_model_path models/bert_weibo_classifier_model.bin \
                                   --epochs_num 10 --batch_size 32 \
                                   --embedding word_pos_seg --encoder transformer --mask fully_visible

下载的bert模型来自谷歌官方 train.tsv: image dev.tsv: image test.tsv: image

三者都是严格按照官方文档进行格式化处理

报错图片: image

之前进行二分类时没有问题。请问进行多分类任务时,是不是需要对run_classifier.py文件进行更改?

ydli-ai commented 2 years ago

把label标签从0开始标试试? 1-6 改为 0-5

Moximixi commented 2 years ago

谢谢楼上大兄弟,可以运行了;但是不知道为什么