Logo
Project Icon

Llama3-Chinese-Chat

基于Llama 3的中英双语优化大语言模型

Llama3Chinese自然语言处理人工智能语言模型Github开源项目

Llama3-Chinese-Chat项目基于Meta-Llama-3-8B-Instruct模型开发，采用ORPO方法优化训练，大幅提升中英双语交互能力。该模型具备角色扮演、工具使用等功能，提供多种版本选择。最新v2.1版本在数学、角色扮演和函数调用方面性能显著提升，训练数据集扩充至10万对。项目同时提供Ollama模型和量化版本，便于快速部署使用。

GithubGithubHuggingfaceHuggingface
介绍相关项目

Llama3-Chinese-Chat

❗️❗️❗️NOTICE: The main branch contains the instructions for Llama3-8B-Chinese-Chat-v2.1. If you want to use or reproduce our Llama3-8B-Chinese-Chat-v1, please refer to the v1 branch; if you want to use or reproduce our Llama3-8B-Chinese-Chat-v2, please refer to the v2 branch.

❗️❗️❗️NOTICE: For optimal performance, we refrain from fine-tuning the model's identity. Thus, inquiries such as "Who are you" or "Who developed you" may yield random responses that are not necessarily accurate.

Updates

Updates for Llama3-8B-Chinese-Chat-v2 [CLICK TO EXPAND]
Updates for Llama3-8B-Chinese-Chat-v1 [CLICK TO EXPAND]

Model Summary

Llama3-8B-Chinese-Chat is an instruction-tuned language model for Chinese & English users with various abilities such as roleplaying & tool-using built upon the Meta-Llama-3-8B-Instruct model.

Developed by: Shenzhi Wang (王慎执) and Yaowei Zheng (郑耀威)

  • License: Llama-3 License
  • Base Model: Meta-Llama-3-8B-Instruct
  • Model Size: 8.03B
  • Context length: 8K

1. Introduction

This is the first model specifically fine-tuned for Chinese & English user through ORPO [1] based on the Meta-Llama-3-8B-Instruct model.

Compared to the original Meta-Llama-3-8B-Instruct model, our Llama3-8B-Chinese-Chat-v1 model significantly reduces the issues of "Chinese questions with English answers" and the mixing of Chinese and English in responses.

Compared to Llama3-8B-Chinese-Chat-v1, our Llama3-8B-Chinese-Chat-v2 model significantly increases the training data size (from 20K to 100K), which introduces great performance enhancement, especially in roleplay, tool using, and math.

[1] Hong, Jiwoo, Noah Lee, and James Thorne. "Reference-free Monolithic Preference Optimization with Odds Ratio." arXiv preprint arXiv:2403.07691 (2024).

Training framework: LLaMA-Factory.

Training details:

  • epochs: 2
  • learning rate: 3e-6
  • learning rate scheduler type: cosine
  • Warmup ratio: 0.1
  • cutoff len (i.e. context length): 8192
  • orpo beta (i.e. $\lambda$ in the ORPO paper): 0.05
  • global batch size: 128
  • fine-tuning type: full parameters
  • optimizer: paged_adamw_32bit

2. Model Download

We provide various versions of our Llama3-8B-Chinese-Chat model, including:

3. Usage

  • Quick use via Ollama

    For the fastest use of our Llama3-8B-Chinese-Chat-v2.1 model, we recommend you use our model via Ollama. Specifically, you can install Ollama here, and then run the following command:

    ollama run wangshenzhi/llama3-8b-chinese-chat-ollama-q4  # to use the Ollama model for our 4bit-quantized GGUF Llama3-8B-Chinese-Chat-v2.1
# or
ollama run wangshenzhi/llama3-8b-chinese-chat-ollama-q8  # to use the Ollama model for our 8bit-quantized GGUF Llama3-8B-Chinese-Chat-v2.1
# or
ollama run wangshenzhi/llama3-8b-chinese-chat-ollama-fp16  # to use the Ollama model for our FP16 GGUF Llama3-8B-Chinese-Chat-v2.1

  • To use the BF16 version of our Llama3-8B-Chinese-Chat model

    You can run the following python script:

    from transformers import AutoTokenizer, AutoModelForCausalLM

model_id = "shenzhi-wang/Llama3-8B-Chinese-Chat"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id, torch_dtype="auto", device_map="auto"
)

messages = [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "写一首诗吧"},
]

input_ids = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True, return_tensors="pt"
).to(model.device)

outputs = model.generate(
    input_ids,
    max_new_tokens=1024,
    do_sample=True,
    temperature=0.6,
    top_p=0.9,
)
response = outputs[0][input_ids.shape[-1]:]
print(tokenizer.decode(response, skip_special_tokens=True))

  • To use the GGUF version of our Llama3-8B-Chinese-Chat model

    First, download the 8bit-quantized GGUF model or f16 GGUF model to your local machine.

    Then, run the following python script:

    from llama_cpp import Llama

model = Llama(
    "/Your/Path/To/Llama3-8B-Chinese-Chat/GGUF/Model",
    verbose=False,
    n_gpu_layers=-1,
)

system_prompt = "You are a helpful assistant."

def generate_reponse(_model, _messages, _max_tokens=8192):
    _output = _model.create_chat_completion(
        _messages,
        stop=["<|eot_id|>", "<|end_of_text|>"],
        max_tokens=_max_tokens,
    )["choices"][0]["message"]["content"]
    return _output

# The following are some examples

messages = [
    {
        "role": "system",
        "content": system_prompt,
    },
    {"role": "user", "content": "写一首诗吧"},
]


print(generate_reponse(_model=model, _messages=messages))

4. Reproduce

To reproduce Llama3-8B-Chinese-Chat-v2.1 (to reproduce Llama3-8B-Chinese-Chat-v1, please refer to this link):

git clone https://github.com/hiyouga/LLaMA-Factory.git
git reset --hard 25aeaae51b6d08a747e222bbcb27e75c4d56a856    # For Llama3-8B-Chinese-Chat-v1: 836ca0558698206bbf4e3b92533ad9f67c9f9864

cd
相关项目
项目侧边栏1项目侧边栏2
推荐项目
Project Cover

豆包MarsCode

豆包 MarsCode 是一款革命性的编程助手，通过AI技术提供代码补全、单测生成、代码解释和智能问答等功能，支持100+编程语言，与主流编辑器无缝集成，显著提升开发效率和代码质量。

Project Cover

AI写歌

Suno AI是一个革命性的AI音乐创作平台，能在短短30秒内帮助用户创作出一首完整的歌曲。无论是寻找创作灵感还是需要快速制作音乐，Suno AI都是音乐爱好者和专业人士的理想选择。

Project Cover

有言AI

有言平台提供一站式AIGC视频创作解决方案，通过智能技术简化视频制作流程。无论是企业宣传还是个人分享，有言都能帮助用户快速、轻松地制作出专业级别的视频内容。

Project Cover

Kimi

Kimi AI助手提供多语言对话支持，能够阅读和理解用户上传的文件内容，解析网页信息，并结合搜索结果为用户提供详尽的答案。无论是日常咨询还是专业问题，Kimi都能以友好、专业的方式提供帮助。

Project Cover

阿里绘蛙

绘蛙是阿里巴巴集团推出的革命性AI电商营销平台。利用尖端人工智能技术，为商家提供一键生成商品图和营销文案的服务，显著提升内容创作效率和营销效果。适用于淘宝、天猫等电商平台，让商品第一时间被种草。

Project Cover

吐司

探索Tensor.Art平台的独特AI模型，免费访问各种图像生成与AI训练工具，从Stable Diffusion等基础模型开始，轻松实现创新图像生成。体验前沿的AI技术，推动个人和企业的创新发展。

Project Cover

SubCat字幕猫

SubCat字幕猫APP是一款创新的视频播放器，它将改变您观看视频的方式！SubCat结合了先进的人工智能技术，为您提供即时视频字幕翻译，无论是本地视频还是网络流媒体，让您轻松享受各种语言的内容。

Project Cover

美间AI

美间AI创意设计平台，利用前沿AI技术，为设计师和营销人员提供一站式设计解决方案。从智能海报到3D效果图，再到文案生成，美间让创意设计更简单、更高效。

Project Cover

AIWritePaper论文写作

AIWritePaper论文写作是一站式AI论文写作辅助工具，简化了选题、文献检索至论文撰写的整个过程。通过简单设定，平台可快速生成高质量论文大纲和全文，配合图表、参考文献等一应俱全，同时提供开题报告和答辩PPT等增值服务，保障数据安全，有效提升写作效率和论文质量。

使用协议隐私政策广告服务
投诉举报邮箱: service@vectorlightyear.com
@2024 懂AI·鲁ICP备2024100362号-6·鲁公网安备37021002001498号