Alignment-Lab-AI/Qwen2.5-Coder-7B-Instruct-132k

1---2license: apache-2.03license_link: https://huggingface.co/Qwen/Qwen2.5-Coder-7B-Instruct/blob/main/LICENSE4language:5- en6base_model:7- Qwen/Qwen2.5-Coder-7B8pipeline_tag: text-generation9library_name: transformers10tags:11- code12- codeqwen13- chat14- qwen15- qwen-coder16---17 18 19# Qwen2.5-Coder-7B-Instruct20 21## Introduction22 23Qwen2.5-Coder is the latest series of Code-Specific Qwen large language models (formerly known as CodeQwen). For Qwen2.5-Coder, we release three base language models and instruction-tuned language models, 1.5, 7 and 32 (coming soon) billion parameters. Qwen2.5-Coder brings the following improvements upon CodeQwen1.5:24 25- Significantly improvements in **code generation**, **code reasoning** and **code fixing**. Base on the strong Qwen2.5, we scale up the training tokens into 5.5 trillion including source code, text-code grounding, Synthetic data, etc. 26- A more comprehensive foundation for real-world applications such as **Code Agents**. Not only enhancing coding capabilities but also maintaining its strengths in mathematics and general competencies.27- **Long-context Support** up to 128K tokens.28 29**This repo contains the instruction-tuned 7B Qwen2.5-Coder model**, which has the following features:30- Type: Causal Language Models31- Training Stage: Pretraining & Post-training32- Architecture: transformers with RoPE, SwiGLU, RMSNorm, and Attention QKV bias33- Number of Parameters: 7.61B34- Number of Paramaters (Non-Embedding): 6.53B35- Number of Layers: 2836- Number of Attention Heads (GQA): 28 for Q and 4 for KV37- Context Length: Full 131,072 tokens38  - Please refer to [this section](#processing-long-texts) for detailed instructions on how to deploy Qwen2.5 for handling long texts.39  40For more details, please refer to our [blog](https://qwenlm.github.io/blog/qwen2.5-coder/), [GitHub](https://github.com/QwenLM/Qwen2.5-Coder), [Documentation](https://qwen.readthedocs.io/en/latest/), [Arxiv](https://arxiv.org/abs/2409.12186).41 42## Requirements43 44The code of Qwen2.5-Coder has been in the latest Hugging face `transformers` and we advise you to use the latest version of `transformers`.45 46With `transformers<4.37.0`, you will encounter the following error:47```48KeyError: 'qwen2'49```50 51## Quickstart52 53Here provides a code snippet with `apply_chat_template` to show you how to load the tokenizer and model and how to generate contents.54 55```python56from transformers import AutoModelForCausalLM, AutoTokenizer57 58model_name = "Qwen/Qwen2.5-Coder-7B-Instruct"59 60model = AutoModelForCausalLM.from_pretrained(61    model_name,62    torch_dtype="auto",63    device_map="auto"64)65tokenizer = AutoTokenizer.from_pretrained(model_name)66 67prompt = "write a quick sort algorithm."68messages = [69    {"role": "system", "content": "You are Qwen, created by Alibaba Cloud. You are a helpful assistant."},70    {"role": "user", "content": prompt}71]72text = tokenizer.apply_chat_template(73    messages,74    tokenize=False,75    add_generation_prompt=True76)77model_inputs = tokenizer([text], return_tensors="pt").to(model.device)78 79generated_ids = model.generate(80    **model_inputs,81    max_new_tokens=51282)83generated_ids = [84    output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)85]86 87response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]88```89 90### Processing Long Texts91 92The current `config.json` is set for context length up to 32,768 tokens.93To handle extensive inputs exceeding 32,768 tokens, we utilize [YaRN](https://arxiv.org/abs/2309.00071), a technique for enhancing model length extrapolation, ensuring optimal performance on lengthy texts.94 95For supported frameworks, you could add the following to `config.json` to enable YaRN:96```json97{98  ...,99  "rope_scaling": {100    "factor": 4.0,101    "original_max_position_embeddings": 32768,102    "type": "yarn"103  }104}105```106 107For deployment, we recommend using vLLM. 108Please refer to our [Documentation](https://qwen.readthedocs.io/en/latest/deployment/vllm.html) for usage if you are not familar with vLLM.109Presently, vLLM only supports static YARN, which means the scaling factor remains constant regardless of input length, **potentially impacting performance on shorter texts**. 110We advise adding the `rope_scaling` configuration only when processing long contexts is required.111 112## Evaluation & Performance113 114Detailed evaluation results are reported in this [📑 blog](https://qwenlm.github.io/blog/qwen2.5-coder/).115 116For requirements on GPU memory and the respective throughput, see results [here](https://qwen.readthedocs.io/en/latest/benchmark/speed_benchmark.html).117 118## Citation119 120If you find our work helpful, feel free to give us a cite.121 122```123@article{hui2024qwen2,124  title={Qwen2. 5-Coder Technical Report},125  author={Hui, Binyuan and Yang, Jian and Cui, Zeyu and Yang, Jiaxi and Liu, Dayiheng and Zhang, Lei and Liu, Tianyu and Zhang, Jiajun and Yu, Bowen and Dang, Kai and others},126  journal={arXiv preprint arXiv:2409.12186},127  year={2024}128}129@article{qwen2,130      title={Qwen2 Technical Report}, 131      author={An Yang and Baosong Yang and Binyuan Hui and Bo Zheng and Bowen Yu and Chang Zhou and Chengpeng Li and Chengyuan Li and Dayiheng Liu and Fei Huang and Guanting Dong and Haoran Wei and Huan Lin and Jialong Tang and Jialin Wang and Jian Yang and Jianhong Tu and Jianwei Zhang and Jianxin Ma and Jin Xu and Jingren Zhou and Jinze Bai and Jinzheng He and Junyang Lin and Kai Dang and Keming Lu and Keqin Chen and Kexin Yang and Mei Li and Mingfeng Xue and Na Ni and Pei Zhang and Peng Wang and Ru Peng and Rui Men and Ruize Gao and Runji Lin and Shijie Wang and Shuai Bai and Sinan Tan and Tianhang Zhu and Tianhao Li and Tianyu Liu and Wenbin Ge and Xiaodong Deng and Xiaohuan Zhou and Xingzhang Ren and Xinyu Zhang and Xipin Wei and Xuancheng Ren and Yang Fan and Yang Yao and Yichang Zhang and Yu Wan and Yunfei Chu and Yuqiong Liu and Zeyu Cui and Zhenru Zhang and Zhihao Fan},132      journal={arXiv preprint arXiv:2407.10671},133      year={2024}134}135```136