Qwen3快速部署 Qwen3-0.6B、Qwen3-8B、Qwen3-14B，Think Deeper

0 Qwen3

今天，通义千问Qwen团队正式开源推出 Qwen3，这是 Qwen 系列大型语言模型的最新成员。最新的Qwen3系列模型具备双模推理能力（深入思考/快速响应）、支持119种语言及方言，并强化了Agent功能与代码执行能力，全面满足复杂问题处理与全球化应用需求。

Github： https://github.com/QwenLM/Qwen3

Blog：https://qwenlm.github.io/zh/blog/qwen3/

模型合集：https://www.modelscope.cn/collections/Qwen3-9743180bdc6b48

b站视频：https://www.bilibili.com/video/BV1spG1zEEYR

1 平台与环境安装

使用GPU平台: https://www.autodl.com/home

PyTorch / 2.3.0 / 3.12(ubuntu22.04) / 12.1

安装transformers、accelerate

source /etc/network_turbo

pip install transformers

pip install accelerate

1 模型下载

Qwen3 模型广场：https://bailian.console.aliyun.com/?tab=model#/model-market

通过魔塔社区下载模型：https://www.modelscope.cn/collections/Qwen3-9743180bdc6b48

选择一个模型Qwen3-0.6B：https://www.modelscope.cn/models/Qwen/Qwen3-0.6B/files

使用SDK下载下载：

开始前安装

source /etc/network_turbo

pip install modelscope

# source /etc/network_turbo
from modelscope import snapshot_download

# 指定模型的下载路径
cache_dir = '/root/autodl-tmp'
# 调用 snapshot_download 函数下载模型
model_dir = snapshot_download('Qwen/Qwen3-0.6B', cache_dir=cache_dir)
# model_dir = snapshot_download('Qwen/Qwen3-8B', cache_dir=cache_dir)
# model_dir = snapshot_download('Qwen/Qwen3-14B', cache_dir=cache_dir)

print(f"模型已下载到: {model_dir}")

或者：

modelscope download --model Qwen/Qwen3-0.6B

mv /root/.cache/modelscope/hub/models/Qwen/ /root/autodl-tmp/Qwen

2 模型测试

from transformers import AutoModelForCausalLM, AutoTokenizer
model_name = "Qwen/Qwen3-0.6B"
# model_name = "Qwen/Qwen3-8B"
# model_name = "Qwen/Qwen3-14B"

# load the tokenizer and the model
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)

# prepare the model input
prompt = "Give me a short introduction to large language models."
messages = [
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
    enable_thinking=True # Switches between thinking and non-thinking modes. Default is True.
)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

# conduct text completion
generated_ids = model.generate(
    **model_inputs,
    max_new_tokens=32768
)
output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist() 

# the result will begin with thinking content in <think></think> tags, followed by the actual response
print(tokenizer.decode(output_ids, skip_special_tokens=True))

enable_thinking=True的结果

enable_thinking=False的结果