52 lines
1.4 KiB
Markdown
52 lines
1.4 KiB
Markdown
|
---
|
|||
|
|
language:
|
|||
|
|
- zh
|
|||
|
|
license: apache-2.0
|
|||
|
|
base_model: Qwen/Qwen2.5-0.5B
|
|||
|
|
pipeline_tag: text-generation
|
|||
|
|
tags:
|
|||
|
|
- 唐诗
|
|||
|
|
- 古诗生成
|
|||
|
|
- chinese-poetry
|
|||
|
|
- qwen2.5
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
# Tangshi｜中文唐诗生成模型
|
|||
|
|
基于Qwen2.5-0.5B微调的古诗专用大模型，擅长自动生成五言/七言绝句、律诗，专为古典诗词创作优化。训练数据为57000首唐诗全参数。
|
|||
|
|
|
|||
|
|
## 仓库信息
|
|||
|
|
Huggingface地址：`Emaoso/Tangshi`
|
|||
|
|
包含两类权重：
|
|||
|
|
1. `model.safetensors`：原生transformers权重，用于Python代码调用
|
|||
|
|
2. `model-ollama.gguf`：GGUF量化权重，用于Ollama本地部署
|
|||
|
|
附带：Ollama一键构建配置 Modelfile
|
|||
|
|
|
|||
|
|
## 一、Python Transformers调用（推荐）
|
|||
|
|
### 1.安装依赖
|
|||
|
|
```bash
|
|||
|
|
pip install torch transformers
|
|||
|
|
|
|||
|
|
|
|||
|
|
====================================
|
|||
|
|
代码示例
|
|||
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM
|
|||
|
|
|
|||
|
|
model_name = "Emaoso/Tangshi"
|
|||
|
|
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
|||
|
|
model = AutoModelForCausalLM.from_pretrained(model_name)
|
|||
|
|
|
|||
|
|
# 写诗指令
|
|||
|
|
prompt = "写一首春日五言绝句"
|
|||
|
|
inputs = tokenizer(prompt, return_tensors="pt")
|
|||
|
|
outputs = model.generate(**inputs, max_new_tokens=80)
|
|||
|
|
result = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
|||
|
|
|
|||
|
|
# 清洗多余注释
|
|||
|
|
import re
|
|||
|
|
result = re.sub(r'（.*|〖.*|见卷.*','',result)
|
|||
|
|
print(result)
|
|||
|
|
|
|||
|
|
======================================
|
|||
|
|
ollama 使用
|
|||
|
|
ollama create tangshi https://huggingface.co/Emaoso/Tangshi/resolve/main/Modelfile
|
|||
|
|
ollama run tangshi
|