- paper :PLOME: Pre-training with Misspelled Knowledge for Chinese Spelling Correction
环境:
- TensorFlow 1.15
- Python3.x
目录结构
- chinesespellingcorrection
- pre_train_src
- finetune_triton
- pinyin_data
- 拼音字典
- stroke_data
- 偏旁字典
- datas
- finance_train.txt
- finance_train.txt
- 保险数据
- train_eval_tagging.py
- fine-tune脚本
- csc_inference.py
- 批量测试脚本
- inference.py
- 同上,这人里面只有function
- flask_server.py
- flask服务测试脚本
- convert_model.py
- tf1.x checkpoint转pb
- export2onnx.sh
- pb转onnx的脚本
- pinyin_data
- triton
- triton pipeline文件 ,plome.onnx文件已删除