--- title: ChineseErrorCorrector2-7B canonical_url: "https://www.modelscope.cn/models/tiannlp/ChineseErrorCorrector2-7B" md_url: "https://www.modelscope.cn/models/tiannlp/ChineseErrorCorrector2-7B.md" repository: tiannlp/ChineseErrorCorrector2-7B last_updated: 2025-08-04 pipeline_tag: text-error-correction tasks: - text-error-correction model_type: - qwen2 architectures: - Qwen2ForCausalLM parameters: 7.6B tensor_type: - BF16 library_name: - transformer - safetensors - pytorch frameworks: - Pytorch inference_backends: - "deploy_task text/emb" - "lmdeploy 0.9.1" - "lmdeploy_turbomind 0.9.1" - "sglang 0.5.2" - "vllm 0.9.2" downloads: 2951 stars: 14 --- # ChineseErrorCorrector2-7B > ChineseErrorCorrector2-7B - tiannlp 在 ModelScope 开源的模型。使用200万纠错数据进行多轮迭代训练，发布了ChineseErrorCorrector2-7B，在 NaCGEC-2023NLPCC官方评测数据集上，超越第一名华为17个点，遥遥领先，推荐使用 tiannlp/ChineseErrorCorrector2-7B 是 ModelScope 魔搭社区上的 7.6B 参数text-error-correction模型，可用 deploy_task text/emb、lmdeploy 0.9.1、lmdeploy_turbomind 0.9.1 部署。 - **Repository**: tiannlp/ChineseErrorCorrector2-7B - **Tasks**: text-error-correction - **Parameters**: 7.6B - **Inference backends**: deploy_task text/emb, lmdeploy 0.9.1, lmdeploy_turbomind 0.9.1, sglang 0.5.2, vllm 0.9.2 - **Downloads**: 2951 - **Stars**: 14 - **Last updated**: 2025-08-04 Source: https://www.modelscope.cn/models/tiannlp/ChineseErrorCorrector2-7B --- # 中文拼写和语法纠错 ----------------- ## 项目地址 https://github.com/TW-NLP/ChineseErrorCorrector 后续会在项目中继续更新纠错模型、量化模型和推理方式。 ## 🔥🔥🔥 新闻 [2025/08/06] 发布[twnlp/ChineseErrorCorrector3-4B](https://huggingface.co/twnlp/ChineseErrorCorrector3-4B) 🎉🎉🎉，泛化性更好，性能更高。 [2025/03/12] 为了方便使用，基于AWQ进行量化，发布 [twnlp/ChineseErrorCorrector2-7B-AWQ](https://huggingface.co/twnlp/ChineseErrorCorrector2-7B-AWQ)，在单张T4(16G)显卡上即可运行😄,transformers推理占用6G显存, [具体详情](https://github.com/TW-NLP/ChineseErrorCorrector) ## 文本纠错评测（拼写错误+语法错误） - 评估工具：ChERRANT [评测工具](https://github.com/HillZhang1999/MuCGEC) - 评估数据：[NaCGEC](https://github.com/masr2000/NaCGEC) - 评估指标：F1-0.5 | Model Name | Model Link | Prec | Rec | F0.5 | |:-----------------|:---------------------------------------------------------------|:-----------|:------------|:-------| | twnlp/ChineseErrorCorrector2-7B | https://huggingface.co/twnlp/ChineseErrorCorrector2-7B | 0.5686 | 0.57 | 0.5689 | | HW_TSC_nlpcc2023_cgec(华为) | 未开源 | 0.5095 | 0.3129 | 0.4526 | | 鱼饼啾啾Plus | 未开源 | 0.5708 | 0.1294 | 0.3394 | | CUHK_SU | 未开源 | 0.3882 | 0.1558 | 0.2990 | | CGEC++ | 未开源 | 0.2414 | 0.0735 | 0.1657 | | zhao_jia | 未开源 | 0.1719 | 0.1478 | 0.1665 | ## 使用 ### transformers ```shell pip install transformers ``` ```shell from transformers import AutoModelForCausalLM, AutoTokenizer model_name = "twnlp/ChineseErrorCorrector2-7B" model = AutoModelForCausalLM.from_pretrained( model_name, torch_dtype="auto", device_map="auto" ) tokenizer = AutoTokenizer.from_pretrained(model_name) prompt = "你是一个文本纠错专家，纠正输入句子中的语法错误，并输出正确的句子，输入句子为：" text_input = "对待每一项工作都要一丝不够。" messages = [ {"role": "user", "content": prompt + text_input} ] text = tokenizer.apply_chat_template( messages, tokenize=False, add_generation_prompt=True ) model_inputs = tokenizer([text], return_tensors="pt").to(model.device) generated_ids = model.generate( **model_inputs, max_new_tokens=512 ) generated_ids = [ output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids) ] response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0] print(response) ``` ### VLLM ```shell pip install transformers pip install vllm==0.3.3 ``` ```shell from transformers import AutoTokenizer from vllm import LLM, SamplingParams # Initialize the tokenizer tokenizer = AutoTokenizer.from_pretrained("twnlp/ChineseErrorCorrector2-7B") # Pass the default decoding hyperparameters of twnlp/ChineseErrorCorrector2-7B # max_tokens is for the maximum length for generation. sampling_params = SamplingParams(seed=42,max_tokens=512) # Input the model name or path. Can be GPTQ or AWQ models. llm = LLM(model="twnlp/ChineseErrorCorrector2-7B") # Prepare your prompts prompt = "对待每一项工作都要一丝不够。" messages = [ {"role": "user", "content": "你是一个文本纠错专家，纠正输入句子中的语法错误，并输出正确的句子，输入句子为："+prompt} ] text = tokenizer.apply_chat_template( messages, tokenize=False, add_generation_prompt=True ) # generate outputs outputs = llm.generate([text], sampling_params) # Print the outputs. for output in outputs: prompt = output.prompt generated_text = output.outputs[0].text print(f"Prompt: {prompt!r}, Generated text: {generated_text!r}") ``` ### VLLM 异步推理 ```shell pip install -r requirements.txt cd ChineseErrorCorrector python main.py ``` ### modelscope ```shell pip install modelscope ``` ```shell from modelscope import AutoModelForCausalLM, AutoTokenizer model_name = "tiannlp/ChineseErrorCorrector2-7B" model = AutoModelForCausalLM.from_pretrained( model_name, torch_dtype="auto", device_map="auto" ) tokenizer = AutoTokenizer.from_pretrained(model_name) prompt = "你是一个文本纠错专家，纠正输入句子中的语法错误，并输出正确的句子，输入句子为：" text_input = "对待每一项工作都要一丝不够。" messages = [ {"role": "user", "content": prompt + text_input} ] text = tokenizer.apply_chat_template( messages, tokenize=False, add_generation_prompt=True ) model_inputs = tokenizer([text], return_tensors="pt").to(model.device) generated_ids = model.generate( **model_inputs, max_new_tokens=512 ) generated_ids = [ output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids) ] response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0] print(response) ``` ## Citation If this work is helpful, please kindly cite as: ```bibtex @inproceedings{wei2024中小学作文语法错误检测, title={中小学作文语法错误检测, 病句改写与流畅性评级的自动化方法研究}, author={Wei, Tian}, booktitle={Proceedings of the 23rd Chinese National Conference on Computational Linguistics (Volume 3: Evaluations)}, pages={278--284}, year={2024} } ``` ## Star History ![Star History Chart](https://api.star-history.com/svg?repos=TW-NLP/ChineseErrorCorrector&type=Date)