---
title: RA-IT-NER-zh-7B
canonical_url: "https://www.modelscope.ai/models/QwenCollection/RA-IT-NER-zh-7B"
md_url: "https://www.modelscope.ai/models/QwenCollection/RA-IT-NER-zh-7B.md"
repository: QwenCollection/RA-IT-NER-zh-7B
last_updated: 2026-11-11
license: cc-by-nc-4.0
pipeline_tag: text-generation
tasks:
  - text-generation
model_type:
  - qwen2
architectures:
  - Qwen2ForCausalLM
parameters: 7.7B
tensor_type:
  - BF16
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - zh
downloads: 521
stars: 0
---

# RA-IT-NER-zh-7B

> RA-IT-NER-zh-7B - An open-source model by QwenCollection on ModelScope. Description: The RA-IT-NER-zh-7B model is trained from Qwen1.5-7B using the proposed Retrieval Augmented Instruction Tuning (RA-IT) approach. This model can be used for Chinese Open NER…

QwenCollection/RA-IT-NER-zh-7B is a 7.7B-parameter text-generation model on ModelScope. licensed under cc-by-nc-4.0.

- **Repository**: QwenCollection/RA-IT-NER-zh-7B
- **License**: cc-by-nc-4.0
- **Tasks**: text-generation
- **Parameters**: 7.7B
- **Downloads**: 521
- **Stars**: 0
- **Last updated**: 2026-11-11

Source: https://www.modelscope.ai/models/QwenCollection/RA-IT-NER-zh-7B

---

---

# RA-IT-NER-zh-7B

**Description**: The RA-IT-NER-zh-7B model is trained from Qwen1.5-7B using the proposed Retrieval Augmented Instruction Tuning (RA-IT) approach. This model can be used for Chinese Open NER with and without RAG. The training data is our constructed [Sky-NER ](https://huggingface.co/datasets/EmmaStrong/Sky-NER), an instruction tuning dataset for Chinese OpenNER. We follow the recipe of [UniversalNER](https://arxiv.org/abs/2308.03279) and use the large-scale [SkyPile Corpus](https://huggingface.co/datasets/Skywork/SkyPile-150B) to construct this dataset. The data was collected by prompting gpt-3.5-turbo-0125 to label entities from passages and provide entity tags. The data collection prompt is as follows:

<div style="background-color: #f6f8fa; padding: 20px; border-radius: 10px; border: 1px solid #e1e4e8; box-shadow: 0 2px 5px rgba(0,0,0,0.1);">
<strong>Instruction:</strong><br/>
给定一段文本，你的任务是抽取所有实体并识别它们的实体类别。输出应为以下JSON格式：[{"实体1": "实体1的类别"}, ...]。</div>

Check our [paper](https://arxiv.org/abs/2406.17305) for more information. Check our [github repo](https://github.com/Emma1066/Retrieval-Augmented-IT-OpenNER) about how to use the model.

## Inference
The template for inference instances is as follows:
<div style="background-color: #f6f8fa; padding: 20px; border-radius: 10px; border: 1px solid #e1e4e8; box-shadow: 0 2px 5px rgba(0,0,0,0.1);">
<strong>Prompting template:</strong><br/>
USER: 以下是一些命名实体识别的例子：<span style="color: #d73a49;">{Fill the NER examples here}</span><br/>
ASSISTANT: 我已读完这些例子。<br/>
USER: 文本：<span style="color: #d73a49;">{Fill the input text here}</span><br/>
ASSISTANT: 我已读完这段文本。<br/>
USER: 文本中属于"<span style="color: #d73a49;">{Fill the entity type here}</span> "的实体有哪些？<br/>
ASSISTANT: <span style="color: #0366d6;">(model's predictions in JSON format)</span><br/>
</div>

Note: 
* The model can conduct inference **with and without** NER examples. If you want to conduct inference without examples, just start from the third line in the above template by directly inputting "文本：{input text}" in the "USER" role.
* Inferences are based on one entity type at a time. For multiple entity types, create separate instances for each type.



## License

This model and its associated data are released under the [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/) license. They are primarily used for research purposes.