cvx-coder: 增强大模型CVX代码能力和CVX问题回答能力

Huggingface模型地址 | 魔搭社区模型地址 | Github地址

简介

cvx-coder 增强了大模型CVX 代码能力和QA能力。它是phi-3在CVX文档, 合成代码, 论坛对话数据上的微调版本。

开始

先从魔搭下载模型：

git clone https://www.modelscope.cn/tommy1235/cvx-coder.git

然后运行下面示例代码

from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
m_path="你的路径/cvx-coder"
model = AutoModelForCausalLM.from_pretrained(
    m_path, 
    device_map="cuda", 
    torch_dtype="auto", 
    trust_remote_code=True, 
)
tokenizer = AutoTokenizer.from_pretrained(m_path)
pipe = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
)
generation_args = {
    "max_new_tokens": 2000,
    "return_full_text": False,
    "temperature": 0,
    "do_sample": False,
}
content='''my problem is not convex, can i use cvx? if not, what should i do, be specific.'''
messages = [
    {"role": "user", "content": content},
]
output = pipe(messages, **generation_args)
print(output[0]['generated_text'])