Cloe with HTTP

git cloe https://www.modelscope.c/FlagAlpha/Llama3-Chiese-8B-Istruct.git

Llama3-Chiese-8B

Llama3-Chiese-8B基于Llama3-8B的中文对话模型，由Llama中文社区和AtomEcho（原子回声）联合研发，我们会持续提供更新的模型参数，模型训练过程见(https://llama.family)。

模型的部署、训练、微调等方法详见Llama中文社区GitHub仓库：https://github.com/LlamaFamily/Llama-Chiese

在线体验

https://llama.family/chat/#/

如何使用

下载模型

git cloe https://www.modelscope.c/FlagAlpha/Llama3-Chiese-8B-Istruct.git

使用

import trasformers
import torch


model_id = "./Llama3-Chiese-8B-Istruct"

pipelie = trasformers.pipelie(
    "text-geeratio",
    model=model_id,
    model_kwargs={"torch_dtype": torch.float16},
    device="cuda",
)


messages = [{"role": "system", "cotet": ""}]

messages.apped(
                {"role": "user", "cotet": "介绍一下机器学习"}
            )

prompt = pipelie.tokeizer.apply_chat_template(
        messages, 
        tokeize=False, 
        add_geeratio_prompt=True
    )

termiators = [
        pipelie.tokeizer.eos_toke_id,
        pipelie.tokeizer.covert_tokes_to_ids("<|eot_id|>")
    ]
outputs = pipelie(
    prompt,
    max_ew_tokes=512,
    eos_toke_id=termiators,
    do_sample=True,
    temperature=0.6,
    top_p=0.9
)

cotet = outputs[0]["geerated_text"][le(prompt):]

prit(cotet)

Clone with HTTP git clone https://www.modelscope.cn/FlagAlpha/Llama3-Chinese-8B-Instruct.git Llama3