Qwen1.5

🤗 Hugging Face   |   🤖 ModelScope   |    📑 Blog    ｜   📖 Documentation
🖥️ Demo   |   💬 WeChat (微信)   |   🫨 Discord  

Visit our Hugging Face or ModelScope organization (click links above), search checkpoints with names starting with Qwen1.5- or visit the Qwen1.5 collection, and you will find all you need! Enjoy!

To learn more about Qwen1.5, feel free to read our documentation [EN|ZH]. Our documentation consists of the following sections:

• Quickstart: the basic usages and demonstrations;
• Inference: the guidance for the inference with transformers, including batch inference, streaming, etc.;
• Run Locally: the instructions for running LLM locally on CPU and GPU, with frameworks like llama.cpp and Ollama;
• Deployment: the demonstration of how to deploy Qwen for large-scale inference with frameworks like vLLM, TGI, etc.;
• Quantization: the practice of quantizing LLMs with GPTQ, AWQ, as well as the guidance for how to make high-quality quantized GGUF files;
• Training: the instructions for post-training, including SFT and RLHF (TODO) with frameworks like Axolotl, LLaMA-Factory, etc.
• Framework: the usage of Qwen with frameworks for applicaition, e.g., RAG, Agent, etc.
• Benchmark: the statistics about inference speed and memory footprint.

Introduction

This time, we upgrade Qwen to Qwen1.5, the beta version of Qwen2. Similar to Qwen, it is still a decoder-only transformer model with SwiGLU activation, RoPE, multi-head attention. At this moment, we have achieved:

• 7 model sizes: 0.5B, 1.8B, 4B, 7B, 14B, and 72B models, plus a 14B (A2.7B) MoE model;
• Significant model quality improvements in chat models;
• Strengthened multilingual capabilities in both base and chat models;
• All models support the context length of 32768 tokens;
• System prompts enabled for all models, which means roleplay is possible.
• No need of trust_remote_code anymore.

We have not integrated GQA and mixture of SWA and full attention in this version and we will add the features in the future version.

News

• 2024.03.28: We released the first MoE model of Qwen: Qwen1.5-MoE-A2.7B! Temporarily, only HF transformers and vLLM support the model. We will soon add the support of llama.cpp, mlx-lm, etc. Check our blog for more information!
• 2024.02.05: We released the Qwen1.5 series.

Performance

Detailed evaluation results are reported in this 📑 blog.

Requirements

• transformers>=4.37.0 for Qwen1.5 dense models.
• For Qwen1.5-MoE models, you should clone transformers and install from source.

Warning

🚨 This is a must because `transformers` integrated Qwen2 codes since `4.37.0`.

Quickstart

🤗 Hugging Face Transformers

Here we show a code snippet to show you how to use the chat model with transformers:

from transformers import AutoModelForCausalLM, AutoTokenizer
device = "cuda" # the device to load the model onto

model = AutoModelForCausalLM.from_pretrained(
    "Qwen/Qwen1.5-72B-Chat",
    torch_dtype="auto",
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen1.5-72B-Chat")

prompt = "Give me a short introduction to large language model."
messages = [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)
model_inputs = tokenizer([text], return_tensors="pt").to(device)

generated_ids = model.generate(
    model_inputs.input_ids,
    max_new_tokens=512
)
generated_ids = [
    output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
]

response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]

For quantized models, we advise you to use the GPTQ and AWQ correspondents, namely Qwen1.5-7B-Chat-GPTQ-Int8, Qwen1.5-7B-Chat-AWQ.

🤖 ModelScope

We strongly advise users especially those in mainland China to use ModelScope. snapshot_download can help you solve issues concerning downloading checkpoints.

💻 Run locally

Ollama

Note

Ollama provides an OpenAI-compatible API, which however does NOT support function calling. For tool use capabilities, consider using Qwen-Agent, which offers a wrapper for function calling over the API.

After installing ollama, you can initiate the ollama service with the following command:

ollama serve
# You need to keep this service running whenever you are using ollama

To pull a model checkpoint and run the model, use the ollama run command. You can specify a model size by adding a suffix to qwen, such as :0.5b, :1.8b, :4b, :7b, :14b, or :72b:

ollama run qwen:4b
# To exit, type "/bye" and press ENTER

You can also access the ollama service via its OpenAI-compatible API. Please note that you need to (1) keep ollama serve running while using the API, and (2) execute ollama run qwen:4b before utilizing this API to ensure that the model checkpoint is prepared.

from openai import OpenAI
client = OpenAI(
    base_url='http://localhost:11434/v1/',
    api_key='ollama',  # required but ignored
)
chat_completion = client.chat.completions.create(
    messages=[
        {
            'role': 'user',
            'content': 'Say this is a test',
        }
    ],
    model='qwen:4b',
)

For additional details, please visit ollama.ai.

llama.cpp

Download our provided GGUF files or create them by yourself, and you can directly use them with the latest llama.cpp with a one-line command:

./main -m <path-to-file> -n 512 --color -i -cml -f prompts/chat-with-qwen.txt

LMStudio

Qwen1.5 has already been supported by lmstudio.ai. You can directly use LMStudio with our GGUF files.

OpenVINO

Qwen1.5 has already been supported by OpenVINO toolkit. You can install and run this chatbot example with Intel CPU, integrated GPU or discrete GPU.

Web UI

Text generation web UI

You can directly use text-generation-webui for creating a web UI demo. If you use GGUF, remember to install the latest wheel of llama.cpp with the support of Qwen1.5.

llamafile

Clone llamafile, run source install, and then create your own llamafile with the GGUF file following the guide here. You are able to run one line of command, say ./qwen.llamafile, to create a demo.

Deployment

Now, Qwen1.5 is supported by multiple inference frameworks. Here we demonstrate the usage of vLLM and SGLang.

Warning

The OpenAI-compatible APIs provided by vLLM and SGLang currently do NOT support function calling. For tool use capabilities, Qwen-Agent provides a wrapper around these APIs to support function calling.

vLLM

We advise you to use vLLM>=0.3.0 to build OpenAI-compatible API service. Start the server with a chat model, e.g. Qwen1.5-7B-Chat:

python -m vllm.entrypoints.openai.api_server --served-model-name Qwen1.5-7B-Chat --model Qwen/Qwen1.5-7B-Chat 

Then use the chat API as demonstrated below:

curl http://localhost:8000/v1/chat/completions \
    -H "Content-Type: application/json" \
    -d '{
    "model": "Qwen1.5-7B-Chat",
    "messages": [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "Tell me something about large language models."}
    ]
    }'

from openai import OpenAI
# Set OpenAI's API key and API base to use vLLM's API server.
openai_api_key = "EMPTY"
openai_api_base = "http://localhost:8000/v1"

client = OpenAI(
    api_key=openai_api_key,
    base_url=openai_api_base,
)

chat_response = client.chat.completions.create(
    model="Qwen1.5-7B-Chat",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Tell me something about large language models."},
    ]
)
print("Chat response:", chat_response)

SGLang

Note

SGLang now does NOT support the Qwen2MoeForCausalLM architecture, thus making qwen1.5-moe-2.7B incompatible.

Please install SGLang from source. Similar to vLLM, you need to launch a server and use OpenAI-compatible API service. Start the server first:

python -m sglang.launch_server --model-path Qwen/Qwen1.5-7B-Chat --port 30000

You can use it in Python as shown below:

from sglang import function, system, user, assistant, gen, set_default_backend, RuntimeEndpoint

@function
def multi_turn_question(s, question_1, question_2):
    s += system("You are a helpful assistant.")
    s += user(question_1)
    s += assistant(gen("answer_1", max_tokens=256))
    s += user(question_2)
    s += assistant(gen("answer_2", max_tokens=256))

set_default_backend(RuntimeEndpoint("http://localhost:30000"))

state = multi_turn_question.run(
    question_1="What is the capital of China?",
    question_2="List two local attractions.",
)

for m in state.messages():
    print(m["role"], ":", m["content"])

print(state["answer_1"])

Finetuning

We advise you to use training frameworks, including Axolotl, Llama-Factory, Swift, etc., to finetune your models with SFT, DPO, PPO, etc.

API

Qwen1.5 models are now deployed on both DashScope and Together. Check this out and have fun with Qwen1.5-72B-Chat!

🐳 Docker

To simplify the deployment process, we provide docker images with pre-built environments: qwenllm/qwen. You only need to install the driver and download model files to launch demos and finetune the model.

docker run --gpus all --ipc=host --network=host --rm --name qwen1.5 -it qwenllm/qwen:1.5-cu121 bash

License Agreement

Check the license of each model inside its HF repo. It is NOT necessary for you to submit a request for commercial usage.

Citation

If you find our work helpful, feel free to give us a cite.

@article{qwen,
  title={Qwen Technical Report},
  author={Jinze Bai and Shuai Bai and Yunfei Chu and Zeyu Cui and Kai Dang and Xiaodong Deng and Yang Fan and Wenbin Ge and Yu Han and Fei Huang and Binyuan Hui and Luo Ji and Mei Li and Junyang Lin and Runji Lin and Dayiheng Liu and Gao Liu and Chengqiang Lu and Keming Lu and Jianxin Ma and Rui Men and Xingzhang Ren and Xuancheng Ren and Chuanqi Tan and Sinan Tan and Jianhong Tu and Peng Wang and Shijie Wang and Wei Wang and Shengguang Wu and Benfeng Xu and Jin Xu and An Yang and Hao Yang and Jian Yang and Shusheng Yang and Yang Yao and Bowen Yu and Hongyi Yuan and Zheng Yuan and Jianwei Zhang and Xingxuan Zhang and Yichang Zhang and Zhenru Zhang and Chang Zhou and Jingren Zhou and Xiaohuan Zhou and Tianhang Zhu},
  journal={arXiv preprint arXiv:2309.16609},
  year={2023}
}

Contact Us

If you are interested to leave a message to either our research team or product team, join our Discord or WeChat groups!