Update README.md
Browse files
README.md
CHANGED
|
@@ -1,8 +1,31 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
-
|
| 4 |
-
-
|
| 5 |
-
|
| 6 |
-
-
|
| 7 |
-
|
| 8 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
+
tags:
|
| 4 |
+
- peft
|
| 5 |
+
- lora
|
| 6 |
+
- Beijing-history
|
| 7 |
+
- knowledge-analysis
|
| 8 |
+
- distilled-model
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# Deepseek-r1-Distill-8b-Beijing-History
|
| 12 |
+
|
| 13 |
+
🤖 模型简介
|
| 14 |
+
基于 Google Colab 微调的轻量化模型,专精于北京历史文化领域知识分析(如胡同演变、皇家建筑、民俗文化等),适用于作为大型任务的中间件或领域知识问答工具。
|
| 15 |
+
|
| 16 |
+
🚀 特性
|
| 17 |
+
- 轻量高效:8B 参数量,推理资源需求低
|
| 18 |
+
- 领域适配:针对北京历史文本优化,理解地名、事件时间线等细节
|
| 19 |
+
- 任务兼容:支持作为 RAG 系统的核心组件或下游任务预处理模块
|
| 20 |
+
|
| 21 |
+
💻 快速使用
|
| 22 |
+
```python
|
| 23 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
| 24 |
+
|
| 25 |
+
model = AutoModelForCausalLM.from_pretrained("JIA244601/beijing112telling")
|
| 26 |
+
tokenizer = AutoTokenizer.from_pretrained("JIA244601/beijing112telling")
|
| 27 |
+
|
| 28 |
+
input_text = "请分析景山在明清两代的功能演变"
|
| 29 |
+
inputs = tokenizer(input_text, return_tensors="pt")
|
| 30 |
+
outputs = model.generate(**inputs, max_new_tokens=200)
|
| 31 |
+
print(tokenizer.decode(outputs[0]))
|