---
title: Tongyi-DeepResearch-30B-A3B-Int4-W4A16
canonical_url: "https://www.modelscope.cn/models/okwinds/Tongyi-DeepResearch-30B-A3B-Int4-W4A16"
md_url: "https://www.modelscope.cn/models/okwinds/Tongyi-DeepResearch-30B-A3B-Int4-W4A16.md"
repository: okwinds/Tongyi-DeepResearch-30B-A3B-Int4-W4A16
chinese_name: "通义深度研究 DeepResearch-30B-A3B-Int4-W4A16 量化高精校准"
last_updated: 2025-09-22
license: apache-2.0
pipeline_tag: text-generation
tasks:
- text-generation
model_type:
- qwen3_moe
architectures:
- Qwen3MoeForCausalLM
base_model:
- iic/Tongyi-DeepResearch-30B-A3B
base_model_relation: quantized
parameters: 4.6B
tensor_type:
- I64
- I32
- BF16
library_name:
- safetensors
- pytorch
frameworks:
- Pytorch
language:
- en
- zh
inference_backends:
- "deploy_task text/emb"
- "lmdeploy_turbomind 0.9.1"
- "sglang 0.5.2"
- "vllm 0.9.2"
downloads: 403
stars: 4
tags:
- Instruct
- Chat
- "Reason Model"
- Quantization
- DeepResearch
---
# Tongyi-DeepResearch-30B-A3B-Int4-W4A16
> Tongyi-DeepResearch-30B-A3B-Int4-W4A16 - okwinds 在 ModelScope 开源的模型。通义深度研究 DeepResearch-30B-A3B-Int4-W4A16 量化高精校准
okwinds/Tongyi-DeepResearch-30B-A3B-Int4-W4A16 是 ModelScope 魔搭社区上的 4.6B 参数text-generation模型，采用 apache-2.0 许可，基于 iic/Tongyi-DeepResearch-30B-A3B 构建，可用 deploy_task text/emb、lmdeploy_turbomind 0.9.1、sglang 0.5.2 部署。
- **Repository**: okwinds/Tongyi-DeepResearch-30B-A3B-Int4-W4A16
- **License**: apache-2.0
- **Tasks**: text-generation
- **Parameters**: 4.6B
- **Base model**: iic/Tongyi-DeepResearch-30B-A3B
- **Inference backends**: deploy_task text/emb, lmdeploy_turbomind 0.9.1, sglang 0.5.2, vllm 0.9.2
- **Tags**: Instruct, Chat, Reason Model, Quantization, DeepResearch
- **Downloads**: 403
- **Stars**: 4
- **Last updated**: 2025-09-22
Source: https://www.modelscope.cn/models/okwinds/Tongyi-DeepResearch-30B-A3B-Int4-W4A16
---
# 通义深度研究 DeepResearch-30B-A3B-Int4-W4A16 量化高精校准
原模型 [iic/Tongyi-DeepResearch-30B-A3B](https://www.modelscope.cn/models/iic/Tongyi-DeepResearch-30B-A3B)
#### 📖 关于量化损失方面的研究，可阅读公众号“觉察流”文章👇
《[Reason Model 的“瘦身计划”：量化技术的得与失](https://mp.weixin.qq.com/s/NMGq4UUkfo8GMix5LHnWCg)》
#### 📖 关于通义DeepResearch方面的论文，可阅读公众号“觉察流”系列文章👇
文章合集：《[【概述】通义DeepResearch开源发布：首个匹敌OpenAI的全栈Web Agent（七篇6.5万字）](https://mp.weixin.qq.com/mp/appmsgalbum?__biz=Mzk2NDA0MzcxNw==&action=getalbum&album_id=4175114324391985174&scene=21)》
- 《[通义 DeepResearch 开源概述](https://mp.weixin.qq.com/s/mSEr3lmylj7-osTnNEhu2Q)》
- 《[WebResearcher：从线性累积到迭代进化，重塑AI研究范式的三大支柱](https://mp.weixin.qq.com/s/JZGZbNVJzPqUlci09tAPlA)》
- 《[ReSum：让AI智能体学会“记笔记”，突破长程推理瓶颈](https://mp.weixin.qq.com/s/w1RHQi5ieOdqlgsRNtGxMQ)》
- 《[WebWeaver：一个模仿人类思维的AI研究员是如何炼成的？](https://mp.weixin.qq.com/s/Nm45ZbwRCFUL3dZVls_qhg)》
- 《[超越规模神话：WebSailor-V2 的数据-环境协同之道](https://mp.weixin.qq.com/s/weB-WcUaMe9z8XRVfQv7yg)》
- 《[从AgentFounder看未来：智能体基础模型的崛起与"持续预训练"新范式](https://mp.weixin.qq.com/s/al7qjD4Ce_W-3HgR1KIZJw)》
- 《[环境即智能：从AgentScaler看下一代AI智能体的进化论](https://mp.weixin.qq.com/s/usiyTnVItzqwVUrryp2hXQ)》
---
#### _作者在此 👇🏻 扫一扫_
---
## 下载
SDK下载
```bash
#安装ModelScope
pip install modelscope
```
```python
#SDK模型下载
from modelscope import snapshot_download
model_dir = snapshot_download('okwinds/DeepResearch-30B-A3B-Int4-W4A16')
```
Git下载
```
#Git模型下载
git clone https://www.modelscope.cn/okwinds/DeepResearch-30B-A3B-Int4-W4A16.git
```
## 模型概述
Tongyi-DeepResearch-30B-A3B-Int4-W4A16 是一个基于 通义千问 Tongyi-DeepResearch-30B-A3B 的 INT4 量化并校准的模型。
- **模型名称:** Tongyi-DeepResearch-30B-A3B-Int4-W4A16
- **模型架构:** Qwen3 MoE
- **权重量化:** INT4
该模型通过将 [iic/Tongyi-DeepResearch-30B-A3B](https://www.modelscope.cn/models/iic/Tongyi-DeepResearch-30B-A3B) 的权重量化为 INT4 数据类型而实现。
量化过程将每个参数从 16bit 减少到 4bit，将模型占用磁盘空间大小，以及推理时加载模型需要的GPU显存空间，减少到了大约为原模型的1/4。只有 transformer 中的 Linear 层权重是量化的，其他层均保持为原模型数据类型，采用了混合精度的计算方式，尽可能减少模型量化后所带来的精度损失。
量化本模型，使用了[AutoGPTQ](https://github.com/AutoGPTQ/AutoGPTQ)，采用 Symmetric group-wise 方式量化（group size : 128）。并在量化过程中，进行了数据校准，以提升模型生成精度。
> 注意：本模型需要 compute capability > 8.0（Ampere、Ada Lovelace、Hopper 架构）的 Nvidia GPU 来支持 INT4 **混合精度**计算。
## 部署推理
#### 推荐使用 vLLM>=0.10.2 ( transformers>=4.56 )
Openai api 兼容模式
```bash
>>> vllm serve "/home/gavin/llm/Tongyi-DeepResearch-30B-A3B-Int4-W4A16" --host 0.0.0.0 --port 8000 --gpu-memory-utilization 0.9 --served-model-name "Tongyi-DeepResearch-30B-A3B-Int4-W4A16"
```
---
# 附录 Tongyi-DeepResearch-30B-A3B 介绍
We present **Tongyi DeepResearch**, an agentic large language model featuring 30 billion total parameters, with only 3 billion activated per token. Developed by Tongyi Lab, the model is specifically designed for **long-horizon, deep information-seeking** tasks. Tongyi-DeepResearch demonstrates state-of-the-art performance across a range of agentic search benchmarks, including BrowserComp-EN, BrowserComp-ZH, GAIA, Humanity's Last Exam, xbench-DeepSearch, and WebWalkerQA.
## Key Features
- ⚙️ **Fully automated synthetic data generation pipeline**: We design a highly scalable data synthesis pipeline, which is fully automatic and empowers agentic pre-training, supervised fine-tuning, and reinforcement learning.
- 🔄 **Large-scale continual pre-training on agentic data**: Leveraging diverse, high-quality agentic interaction data to extend model capabilities, maintain freshness, and strengthen reasoning performance.
- 🔁 **End-to-end reinforcement learning**: We employ a strictly on-policy RL approach based on a customized Group Relative Policy Optimization framework, with token-level policy gradients, leave-one-out advantage estimation, and selective filtering of negative samples to stabilize training in a non‑stationary environment.
- 🤖 **Agent Inference Paradigm Compatibility**: At inference, Tongyi-DeepResearch is compatible with two inference paradigms: ReAct, for rigorously evaluating the model's core intrinsic abilities, and an IterResearch-based 'Heavy' mode, which uses a test-time scaling strategy to unlock the model's maximum performance ceiling.
## Download
You can download the model then run the inference scipts in https://github.com/Alibaba-NLP/DeepResearch.
```bibtex
@misc{tongyidr,
author={Tongyi DeepResearch Team},
title={Tongyi-DeepResearch},
year={2025},
howpublished={\url{https://github.com/Alibaba-NLP/DeepResearch}}
}
```