---
title: Qwen2.5-7b-ins-merged-awq
canonical_url: "https://www.modelscope.cn/models/xiaohaiqu/Qwen2.5-7b-ins-merged-awq"
md_url: "https://www.modelscope.cn/models/xiaohaiqu/Qwen2.5-7b-ins-merged-awq.md"
repository: xiaohaiqu/Qwen2.5-7b-ins-merged-awq
last_updated: 2026-01-02
license: "Apache License 2.0"
pipeline_tag: text-generation
tasks:
  - text-generation
  - question-answering
model_type:
  - qwen2
architectures:
  - Qwen2ForCausalLM
base_model:
  - Qwen/Qwen2.5-7B-Instruct
base_model_relation: finetune
parameters: 2.0B
tensor_type:
  - F16
  - I32
library_name:
  - transformer
  - lora
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - zh
inference_backends:
  - "deploy_task text/emb"
  - "lmdeploy 0.9.1"
  - "lmdeploy_turbomind 0.9.1"
  - "sglang 0.5.2"
  - "vllm 0.9.2"
downloads: 32
stars: 3
tags:
  - gpu-programming
  - cuda
  - qwen2.5
  - lora
  - fine-tuned
---

# Qwen2.5-7b-ins-merged-awq

> Qwen2.5-7b-ins-merged-awq - xiaohaiqu 在 ModelScope 开源的模型。Qwen2.5-7B-Instruct GPU编程专家

xiaohaiqu/Qwen2.5-7b-ins-merged-awq 是 ModelScope 魔搭社区上的 2.0B 参数text-generation、question-answering模型，采用 Apache License 2.0 许可，基于 Qwen/Qwen2.5-7B-Instruct 构建，可用 deploy_task text/emb、lmdeploy 0.9.1、lmdeploy_turbomind 0.9.1 部署。

- **Repository**: xiaohaiqu/Qwen2.5-7b-ins-merged-awq
- **License**: Apache License 2.0
- **Tasks**: text-generation, question-answering
- **Parameters**: 2.0B
- **Base model**: Qwen/Qwen2.5-7B-Instruct
- **Inference backends**: deploy_task text/emb, lmdeploy 0.9.1, lmdeploy_turbomind 0.9.1, sglang 0.5.2, vllm 0.9.2
- **Tags**: gpu-programming, cuda, qwen2.5, lora, fine-tuned
- **Downloads**: 32
- **Stars**: 3
- **Last updated**: 2026-01-02

Source: https://www.modelscope.cn/models/xiaohaiqu/Qwen2.5-7b-ins-merged-awq

---

# Qwen2.5-7B-Instruct GPU编程专家

基于 Qwen2.5-7B-Instruct 微调和 awq 量化的 GPU 编程问答模型，专注于 CUDA/PTX/GPU 架构相关知识。

适用于中国科学院大学 2025 年秋季《GPU 架构与编程》课程大作业一

## 模型信息

| 属性 | 值 |
|------|-----|
| 基座模型 | Qwen2.5-7B-Instruct |
| 微调方法 | LoRA (rank=64, alpha=128) |
| 精度 | FP16 |
| 训练数据 | GPU编程 Q&A 数据集 |

## 使用方法

### SDK 下载

```bash
pip install modelscope
```

```python
from modelscope import snapshot_download
model_dir = snapshot_download('xiaohaiqu/Qwen2.5-7b-ins-merged-awq')
```

### Git 下载

```bash
git clone https://www.modelscope.cn/xiaohaiqu/Qwen2.5-7b-ins-merged-awq.git
```

### 推理示例

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained(
    "xiaohaiqu/Qwen2.5-7b-ins-merged-awq",
    torch_dtype="auto",
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained("xiaohaiqu/Qwen2.5-7b-ins-merged-awq")

messages = [
    {"role": "system", "content": "你是GPU编程专家，精通CUDA/PTX/GPU架构。"},
    {"role": "user", "content": "什么是 shared memory？"}
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer([text], return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```

## 适用场景

- CUDA 编程概念问答
- GPU 架构知识查询
- 并行计算原理解释
- Triton/TileLang 相关问题

---

**欢迎 Star 和 Follow！**

## 许可证

Apache License 2.0
