Skip to content

heli-dawnlab703/pytorch_bert_multi_classification

Repository files navigation

启动笔记 -- > 2023-9-28

  1. 可以运行run.sh
  2. checkpoints 放置预训练好的模型
  3. 需要从huggingface中下载模型, 下载对应模型 中文的对应bert_base_chiese, 并放置model_hub目录中! link

error

  1. GH001: Large files detected. You may want to try Git Large File Storage - https://git-lfs.git 解决方法

调优

bert_base_chinese

--lr=3e-5 --model_name rank --other_lr=3e-4 --train_batch_size=64 --eval_batch_size=64 --train_epochs=11 --lr=3e-6 --model_name rank --other_lr=3e-4 --train_batch_size=32 --eval_batch_size=32 --train_epochs=11 --bert_dir ./model_hub/bert-base-chinese

roberta-wwm-ext-large

--lr=3e-5 --model_name rank --other_lr=3e-4 --train_batch_size=64 --eval_batch_size=64 --train_epochs=11 --lr=3e-6 --model_name rank --other_lr=3e-4 --train_batch_size=16 --eval_batch_size=16 --train_epochs=11 --bert_dir ./model_hub/roberta-wwm-ext-large 分类: 不太行 阈值为0.5 --lr=3e-5 --model_name rank --other_lr=3e-4 --train_batch_size=64 --eval_batch_size=64 --train_epochs=11 【test】 loss:607.445513 accuracy:0.8530 micro_f1:0.8530 macro_f1:0.8180

改版后

使用高斯进行分类 效果达到0.88%

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published