77 lines
1.7 KiB
Markdown
77 lines
1.7 KiB
Markdown
---
|
||
frameworks:
|
||
- Pytorch
|
||
license: Apache License 2.0
|
||
tasks:
|
||
- text-classification
|
||
|
||
#model-type:
|
||
##如 gpt、phi、llama、chatglm、baichuan 等
|
||
- llama
|
||
|
||
#domain:
|
||
##如 nlp、cv、audio、multi-modal
|
||
- nlp
|
||
|
||
#language:
|
||
##语言代码列表 https://help.aliyun.com/document_detail/215387.html?spm=a2c4g.11186623.0.0.9f8d7467kni6Aa
|
||
- cn
|
||
- zh
|
||
|
||
#metrics:
|
||
##如 CIDEr、Blue、ROUGE 等
|
||
#- CIDEr
|
||
|
||
#tags:
|
||
##各种自定义，包括 pretrained、fine-tuned、instruction-tuned、RL-tuned 等训练方法和其他
|
||
- fine-tuned
|
||
|
||
#tools:
|
||
##如 vllm、fastchat、llamacpp、AdaSeq 等
|
||
- vllm
|
||
|
||
---
|
||
# Unichat-llama3-Chinese-8B
|
||
|
||
|
||
## 介绍
|
||
* 中国联通AI创新中心发布llama3.1中文指令微调模型，全参数微调
|
||
* 基础模型 [**Meta-Llama-3.1-8B-Instruct**](https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct)
|
||
|
||
|
||
### 📊 数据
|
||
- 高质量指令数据，覆盖多个领域和行业，为模型训练提供充足的数据支持
|
||
- 微调指令数据经过严格的人工筛查，保证优质的指令数据用于模型微调
|
||
|
||
|
||
```python
|
||
import transformers
|
||
import torch
|
||
|
||
model_id = "UnicomAI/Unichat-llama3.1-Chinese-8B"
|
||
|
||
pipeline = transformers.pipeline(
|
||
"text-generation",
|
||
model=model_id,
|
||
model_kwargs={"torch_dtype": torch.bfloat16},
|
||
device_map="auto",
|
||
)
|
||
|
||
messages = [
|
||
{"role": "system", "content": "You are a helpful assistant"},
|
||
{"role": "user", "content": "你是谁?"},
|
||
]
|
||
|
||
outputs = pipeline(
|
||
messages,
|
||
max_new_tokens=1024,
|
||
do_sample=False,
|
||
repetition_penalty=1.1,
|
||
)
|
||
print(outputs[0]["generated_text"][-1])
|
||
```
|
||
|
||
## 资源
|
||
更多模型,数据集和训练相关细节请参考:
|
||
* Github：[**Unichat-llama3-Chinese**](https://github.com/UnicomAI/Unichat-llama3-Chinese)
|
||
|