HYY

KO | EN

gitlite — search

#go #golang #android #machine-learning #rust #deep-learning #docker #java #windows #cli #vue #security #mcp #openai

(★ 48)

Building a Large Language Model Agent to Become an Expert in the Field of Ceramics.

HYY 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# HYY
# 🏺ChatCAS: Multimodal Ceramic Multi-Agent Studio (Consultation • Analysis • Generation)
The original repository was at [https://github.com/HYY](https://github.com/HYY), and has now been moved to this repository.

![CeramicGPT overview](https://raw.githubusercontent.com/HanYongyi/HYY/main/image.png)

**🔬 Agent Construction Framework for the Ceramics Domain: Application Demonstrations in Ceramic Knowledge Question-Answering and Image Analysis & Generation**  
**📅 Estimated completion time of open source project: early 2026**

---

## 📖 Project Overview
This project aims to develop the first large language model (LLM) system specifically for the ceramic field. This system integrates rich multimodal knowledge of ceramics, multimodal evaluation datasets, and a collaborative multi-agent framework, all of which are designed to enhance expert consultation, image analysis, and creative decision-making throughout the entire ceramic design and production process.

We propose the following key modules:：
- 📊 **EvalCera**：The first multimodal evaluation dataset in the ceramics domain
- 🏺 **CeramicGPT**：The first large language model tailored for the ceramics field, capable of intelligently handling knowledge-based Q&A in the ceramics field.
- 🤖 **ChatCAS**：Driven by CeramicGPT and GPT-4o. It categorizes tasks related to the various stages of a ceramic studio, completing collaborative work through role-playing and task goal consensus. ChatCAS can effectively handle comprehensive tasks in the ceramics field, including Q&A, image analysis, and generation.
The project is dedicated to promoting the deep integration of artificial intelligence and ceramics, and supporting more scientific research and industrial application explorations.

The project is dedicated to promoting the deep integration of artificial intelligence and ceramics, and supporting more scientific research and application exploration.

---

## 📌 Commercialization Notice

The project has officially entered the commercialization phase.  
Due to intellectual property protection, compliance requirements, and ongoing industrial applications, the majority of the core assets now serve commercial use cases — including:

- Model weight files  
- Full agent modules and orchestration logic  
- Most datasets and production-grade pipelines  
- The latest mini-program versions and deployment architecture

These components have been transferred to **Tongchuan Damenli Ceramics Co., Ltd. (铜川市大门里瓷业公司)** for real-world deployment and are therefore **no longer available for full open-source release**. 🔒

Accordingly, the original open-source roadmap has been discontinued. ✅

That said, out of goodwill and to support learning, reproducibility, and community exploration, we are keeping **a limited set of materials publicly available**, including:

- A **partial dataset subset** (non-sensitive / compliance-reviewed)  
- The **first-generation (prototype) version** of the mini-program

We hope these retained resources can still be helpful for academic readers and developers who want to understand the early-stage design choices and implementation ideas. 🙏

---

### Academic Trial Access 🎓

To promote academic exchange, we are willing to provide **trial access to the ChatCAS system** for qualified academic collaborators.

Interested researchers may contact us via WeChat:

> +86 18511601531 💬

After adding the account, please leave a message indicating your request for a system trial.

---

### Research & Commercial Cooperation 🤝

We welcome cooperation with qualified third-party research institutions or academic teams through formal written agreements.

Under the premise of:

- Data privacy protection  
- Research ethics compliance  
- Intellectual property protection

we are willing to provide **authorized access to the full ChatCAS system (100%)** for eligible partners. ✅

For commercial collaboration inquiries, please contact us directly to discuss further arrangements. 📩

---------------------------------------------------------------------------------------------------------

## 🚀 Model Availability

The **CeramicGPT** model and related resources are available on [Hugging Face](https://huggingface.co/HanYongyi/CeramicGPT).  
---

## 💻 Mini Program: Porcelain Language Intelligent Q&A and Image Analysis Mini-Program

To make the ChatCAS system more accessible and intuitive, we have developed a WeChat Mini Program named “瓷语”, which integrates large model–based ceramic knowledge Q&A and multimodal image analysis.
Due to commercial collaboration confidentiality, the source code of this mini program is partly open-sourced in this repository under the folder /MiniProgram/.

Key Features:

Intelligent Q&A via text or voice input, supporting multi-turn contextual dialogue.

Image-based analysis of ceramic artifacts, identifying shape, texture, and style patterns.

Integration with ChatCAS for creative interpretation and design assistance.

Lightweight WeChat-native UI for smooth interaction and visualization.

<div align="center">
  <img src="./miniapp1.png" width="20%" alt="瓷语小程序界面1">
  <img src="./miniapp2.png" width="20%" alt="瓷语小程序界面2">
</div>
## 📁 Directory Structure

```bash
CeramicGPT/
Mini Program/
├── data/                       # Knowledge Graph and Evaluation Set (Coming soon)
├── models/                     # Fine-tuning model related content (open source planned)
├── agents/                     # Multi-agent framework code
├── docs/                       # Project documentation and experimental details
└── README.md                   # Project description
```
## 🔄 Reproducibility and Data Standards

To ensure transparency and reproducibility, we have made the following resources publicly available in this repository:

- **Reproducibility elements**:  
  - Data split scripts for training/validation/test sets
  - Fixed random seed configurations to guarantee consistent results
  - Detailed hardware specifications (GPU/CPU environment), including tested environments
  - Minimum GPU requirement: 24GB VRAM (e.g., NVIDIA RTX 4090 or higher)

These resources enable independent verification of all reported results in the paper.

- **Data copyright and cleaning standards**:  
  - Image data were mainly collected from **museum collections** and **Baidu Baike**.  
  - Copyright attribution has been clarified, and only publicly shareable resources are included.  
  - A standardized data-cleaning process (removal of duplicates, low-quality or unclear samples) has been applied to ensure the reliability and fairness of the dataset.

👉 For details, please refer to the (./docs) folder and experiment notes in this repository.
---

## 📜 License

The project will be open source under the Apache 2.0 License or MIT License for academic and commercial use. The final license will be specified at the time of the open source release.
---

## 📬 Contact

If you have the intention of cooperation or exchange, please contact us through the following ways：

- 📧 Email: 1399780501@qq.com(Yongyi Han); 1241917171@qq.com (Diandong Liu)

> “让古老的瓷火与新生的大模型在同一炉温中交融”

HYY

HYY 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

HYY

HYY 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README