Release Kuwa v0.3.4

Feature Updates

1. Consolidated offline English documentation #39
2. Support for direct integration with third-party APIs. The default model now includes Groq Llama3.1 70B.
3. A Pre-built Docker Image is provided for faster installation.
4. The Docker version directly supports CUDA, eliminating the need for repeated container installation.
5. Added Kuwa Javascript Library for calling Multi-chat API.
6. Ability to directly edit Bot settings by clicking the Bot image within the chat room.
7. RAG supports caching Embedding models and Vector DB.
8. SearchQA facilitates integration with third-party search engines.
9. DocQA/WebQA offers a Failback mode, allowing direct interaction with an LLM.
10. Pipes can now specify parameters from the prompt.
11. Added Media Converter for editing and splicing videos or audio.
12. The default Gemini model has been updated to Gemini 1.5 Flash.

Bug Fixes

1. Docker WARN message "FromAsCasing: 'as' and 'FROM' keywords' casing do not match" #38
2. The Docker Version Executor diagram disappears
3. Chinese display garbled on Windows version
4. The Chinese file path is not supported when creating Vector DB on Windows version
5. The file name is too long and exceeds the path length limit on Windows version
6. ChatGPT executor does not follow custom context_window

Warning

The Docker version defaults to using pre-build Docker images, if you need to build from the source, please refer to this document.

Contributors

• Update Dockerfile by @cclljj in #38
• [docs] add: docs directory by @wcwutw in #39

Full Changelog: v0.3.3...v0.3.4

---

We welcome your feedback after experiencing it. If you encounter any difficulties, please feel free to contact us through the community or other channels.

Kuwa official website: https://kuwaai.tw/
Kuwa introduction: https://kuwaai.tw/blog/kuwa-os-intro

Release Kuwa v0.3.3

Feature Updates

1. Added Pipe executor, which can execute programs (tools) within a specified directory, such as directly executing Python programs output by models via Python interpreter
2. Provided Calculator, Iconv and Python example tools that can be called via Pipe executor
3. Added Uploader executor to allow users to upload files to a specified directory, including tools, RAG knowledge bases, or website components
4. Supported Bot export and import, allowing export of Bot name, description, icon, and Modelfile as a single Bot file, similar to an application configuration file; installation can automatically import default Bot files
5. Allows users to choose the sorting method for Bots in chat rooms
6. Supported Bot icon replacement
7. Added Kuwa API server compatible with OpenAI API
8. Provided default examples for connecting to cloud multimodal APIs: gpt-4o-mini-vision, DALL-E, Gemini pro 1.5 vision
9. Supported setting the upper limit of uploaded files via Web interface
10. Supported installation and execution in environments with Web proxy within enterprises
11. Supported acceleration of model inference using Intel GPU
12. Added automatic installation and update scripts for Docker version, thanks to @wcwutw
13. RAG Toolchain default Embedding model replaced with Microsoft's intfloat/multilingual-e5-small model, licensed under MIT
14. RAG (DocQA, WebQA, SearchQA, DB QA) added display_hide_ref_content, retriever_ttl_sec parameters
15. Increased support for downloaded tools' default models, including Meta Llama 3.1 8B with Function calling and lightweight Google Gemma 2 2B

Bug Fixes

1. #21: Docker version does not generate https:// links after a reverse proxy, thanks to @Phate334
2. #23: Two-minute timeout issue, thanks to @x85432
3. #24: Modelfile parsing issue
4. #25: Importing prompts does not apply Modelfile
5. windows\src\tee.bat is misjudged as a virus
6. RAG reference data does not display original file names
7. Updated Windows version dependency download link

Warning

1. Windows version may encounter 502 Bad Gateway error when upgrading from old version, please remove C:\kuwa\GenAI OS\windows\packages\nginx-1.24.0\conf\nginx.conf and re-run build.bat once
2. Pipe executor can execute any program and currently lacks Chroot, please understand related risks before providing external services

New Contributors

• @Phate334 made their first contribution in #21
• @vivianh38 made their first contribution in #22
• @wcwutw made their first contribution in #26
• @wuchtw made their first contribution in #28
• @x85432 made their first contribution in #33

Full Changelog: v0.3.2...v0.3.3

---

We welcome your feedback after experiencing it. If you encounter any difficulties, please feel free to contact us through the community or other channels.

Kuwa official website: https://kuwaai.tw/
Kuwa introduction: https://kuwaai.tw/blog/kuwa-os-intro

Release Kuwa v0.3.2

Feature Updates

1. Customized Bot Permissions: Configure the Bot's readable and executable permissions at system, community, group, and individual levels
2. Customized Upload File Policy: Admin can set maximum upload file size and allowed file types
3. Tool Samples: Added samples for Copycat, token counter, etc.
4. Pre-defined Model Profiles: Provided profiles for LLaVA and other fine-tuned models
5. UX Optimization: Beautified icons and chat lists
6. Updated Default Models: ChatGPT Executor is connected to GPT-4o by default, Gemini Executor is connected to Gemini 1.5 pro by default

Bug Fixes

1. File name with whitespace parsing issue when uploading
2. Language is not saved after logout
3. Dependency issue of Llamacpp Executor
4. Color and line breaks not supported in Windows version logs
5. The first message in the group chat is always sent even using multi-chat single-turn Q&A
6. Windows version DocQA default parameters may exceed the context window

Full Changelog: v0.3.1...v0.3.2

---

Welcome to provide feedback after experience. Please contact us via the community or other channels if you encounter any issues.

Kuwa Official Website: https://kuwaai.org/
Kuwa Introduction: https://kuwaai.org/blog/kuwa-os-intro

Release Kuwa v0.3.1

Hi everyone, Kuwa v0.3.1 is out, and this update mainly focuses on multimodal input and output, which now supports both speech and images. Combined with the previously launched Bot system and group chat functions, this allows for practical functions such as meeting summaries, speech summaries, simple image generation, and image editing:

1. Supports the Whisper speech-to-text model, which can output transcripts from uploaded audio files, and features multi-speaker recognition and timestamps.
2. Supports the Stable Diffusion image generation model, which can generate images from text input or modify uploaded images based on user instructions.
3. Huggingface executor supports integration with vision-language models such as Phi-3-Vision and LLaVA.
4. RAG supports direct parameter adjustment through the Web UI and Modelfile, simplifying the calibration process.
5. RAG supports displaying original documents and cited passages, making it easier to review search results and identify hallucinations.
6. Supports importing pre-built RAG vector databases, facilitating knowledge sharing across different systems.
7. Simplified selection of various open models during installation.
8. Multi-chat Web UI supports direct export of chat records in PDF, Doc/ODT formats.
9. Multi-chat Web UI supports Modelfile syntax highlighting, making it easy to edit Modelfiles.
10. Kernel API supports passing website language, allowing the Executor to customize based on user language.
11. The Executor removes the default System prompt to avoid compromising model performance.

Here are the detailed user guide documents:

1. Whisper Speech-to-Text Model User Guide (including speaker recognition): https://kuwaai.org/blog/whisper-tutorial
2. Stable Diffusion Image Generation Model User Guide: https://kuwaai.org/blog/diffusion-tutorial
3. Vision and Language Model Integration Tutorial: https://kuwaai.org/blog/vlm-tutorial

We welcome your feedback after trying out the new version, and please feel free to contact us through the community or other channels if you encounter any difficulties.

Official Kuwa website: https://kuwaai.org/

---

Introduction to Kuwa GenAI OS

Kuwa GenAI OS is a free, open, secure, and privacy-focused open-source system that provides a user-friendly interface for generative AI and a new-generation generative AI orchestrator system that supports rapid development of LLM applications.
Kuwa provides an end-to-end solution for multilingual and multi-model development and deployment, empowering individuals and industries to use generative AI on local laptops, servers or the cloud, develop applications, or open stores and provide services externally.
Here is a brief description of version v0.3.1:

Usage Environment

1. Supports multiple operating systems including Windows, Linux, and MacOS, and provides easy installation and software update tools, such as a single installation executable for Windows, an automatic installation script for Linux, a Docker startup script, and a pre-installed VM virtual machine.
2. Supports a variety of hardware environments, from Raspberry Pi, laptops, personal computers, and on-premises servers to virtual hosts, public and private clouds, with or without GPU accelerators.

User Interface

1. The integrated interface can select any model, knowledge base, or GenAI application, and combine them to create single or group chat rooms.
2. The chat room can be self-directed, citing dialogue, specifying group chat or direct private chat, switching between continuous Q&A mode or single-question Q&A mode
3. Controllable crossings at any time, import prompt scripts or upload files, you can also export complete chat room conversation scripts, directly output files in formats such as PDF, Doc/ODT, plain text, or share web pages
4. Supports text, image generation, speech, and visual recognition multimodal language models, and can highlight syntax such as programming and Markdown, or quickly use system gadgets.

Development Interface

1. Users can skip coding by connecting existing models, knowledge bases, or Bot applications, adjusting system prompts and parameters, presetting scenarios, or creating prompt templates to create personalized or more powerful GenAI applications.
2. Users can create their own knowledge base by simple drag and drop, or import existing vector databases, and can use multiple knowledge bases for GenAI applications at the same time.
3. Users can create and maintain their own shared app Store, and users can also share bot apps
4. The Kuwa extension model and RAG advanced functions can be adjusted and enabled through the Ollama modelfile.

Deployment Interface

1. Supports multiple languages, can customize the interface and messages, and directly provide services for external deployment.
2. Existing accounts can be connected or registered with an invitation code. When the password is forgotten, it can be reset with Email.
3. System settings can modify system announcements, terms of service, warnings, etc., or perform group permission management, user management, model management, etc.
4. The dashboard supports feedback management, system log management, security and privacy management, message query, etc.

Development Environment

1. Integrates a variety of open-source generative AI tools, including Faiss, HuggingFace, Langchain, llama.cpp, Ollama, vLLM, and various Embedding and Transformer-related packages. Developers can download, connect, and develop various multimodal LLMs and applications.
2. RAG Toolchain includes multiple search-augmented generation application tools such as DBQA, DocumentQA, WebQA, and SearchQA, which can be connected with search engines and automatic crawlers, or integrated with existing corporate databases or systems, facilitating the development of advanced customized applications.
3. Open source allows developers to create their own custom systems based on their own needs.

Full Changelog: v0.3.0...v0.3.1

Release Kuwa v0.3.0

Hello everyone, we have released the formal version of kuwa-v0.3.0 with the feedback and suggestions from the community, hoping that it could satisfy everyone's needs.

What has been changed

The major differences from the previous version kuwa-v0.2.1 are that we have added and enhanced the functions of Bot, Store, RAG toolchain, and system updates, and a new chat and group chat integrated interface:

1. Bot enables users to create Bot applications without code, and they can adjust System prompt, preset chat records, and User prompt prefixes and suffixes to realize various functions such as role-playing, executing specific tasks, or using the Ollama Modelfile to create more powerful applications;
2. Store allows users to build and maintain shared Bot application stores on their own, and users can also share Bots with each other;
3. The RAG toolchain allows users to easily drag and drop local file folders to create their own vector databases, and then use the existing DBQA function to perform Q&A through multiple databases;
4. Not only does the new integrated interface directly support group chat and single-model chat, but you can also import Prompt Sets or upload files at any time, and also be used for related RAGs;
5. A SearchQA was added to the Windows version, which can be connected to Google search to perform Q&A on organizational websites;
6. A Docker startup script was added to simplify the startup method of the Docker version (thanks to @cclljj );
7. Executor can be directly connected to Ollama to use the models and applications supported by Ollama;
8. You can use update.bat to quickly update to the latest released version without having to re-download the .exe installation file

Complete changelog: v0.2.1-taide...v0.3.0

Related tutorial documents

We also provide some of Kuwa's tutorial documents:

1. DB QA/RAG setup tutorial: https://kuwaai.org/blog/dbqa-setup
2. Bot setup tutorial: https://kuwaai.org/blog/bot-system-guide
3. Docker startup tutorial: https://github.com/kuwaai/genai-os/blob/main/docker/README.md
4. SearchQA setup tutorial: https://kuwaai.org/blog/search-qa-setup
5. Gemini API Key application tutorial: https://kuwaai.org/blog/apply-gemini

You are welcome to give us feedback after trying it out and feel free to contact us through the community or other channels if you encounter any difficulties.
Kuwa official website: https://kuwaai.org/

---

大家好，經過社群的回饋與建議，我們推出 kuwa-v0.3.0 正式版，以期更能滿足大家的需求。

改了什麼

與前一版 kuwa-v0.2.1 的主要差異為新增並強化了 Bot、Store、RAG toolchain和系統更新等功能，以及新的聊天與群聊整合介面：

1. Bot 讓使用者可以no code建立Bot應用，可調整System prompt、預置聊天紀錄、User prompt前後綴來實現如角色扮演、執行特定任務等不同功能，或利用 Ollama Modelfile 來打造更強大的應用；
2. Store 讓使用者可以自行建置及維護共用的Bot應用商店，使用者間也可以共享Bot；
3. RAG toolchain 讓使用者可以將本機文件資料夾透過簡易拖拉的方式建立成自有的向量資料庫，不限一個，再透過既有的 DBQA 功能進行問答；
4. 新的整合介面不僅直接支援群聊與單模型聊天，隨時可以匯入Prompt Sets或上傳檔案，也可以用於相關RAG；
5. Windows 版新增 SearchQA，可透過串接 Google search 進行組織網站問答；
6. 新增 Docker 啟動腳本，簡化 Docker 版啟動方式 (感謝 @cclljj )；
7. Executor 可以直接串接 Ollama，藉以使用 Ollama 支援的模型及應用；
8. 可以透過 update.bat 來快速更新至最新釋出的版本，不必重新下載 .exe 安裝檔

完整變更紀錄: v0.2.1-taide...v0.3.0

相關教學文件

另外也提供幾份Kuwa的教學文件：

1. DB QA/RAG 建置教學: https://kuwaai.org/zh-Hant/blog/dbqa-setup
2. Bot 建置教學: https://kuwaai.org/zh-Hant/blog/bot-system-guide
3. Docker 啟動教學: https://github.com/kuwaai/genai-os/blob/main/docker/README_TW.md
4. SearchQA 建置教學: https://kuwaai.org/zh-Hant/blog/search-qa-setup
5. Gemini API Key 申請教學: https://kuwaai.org/zh-Hant/blog/apply-gemini

歡迎各位體驗之後給予回饋，遇到任何困難也歡迎透過社群或是其他管道與我們聯絡。
Kuwa官網: https://kuwaai.org/zh-Hant/

Release Kuwa v0.3.0-beta1

Hello, friends of the community,

As planned, we have launched the kuwa-v0.3.0-beta1 version for you to experience in advance 😁

What's Changed

This version mainly adds new functions such as Bot, Store, RAG toolchain, as well as new integrated chat and group chat interface:

1. Bot allows users to create Bot applications without code, and the System prompt can be adjusted to achieve different functions such as role-playing and performing specific tasks. The model supports partial Ollama model file settings;
2. Store allows users to independently build and maintain a shared Bot application store, and users can also share the Bot;
3. RAG toolchain allows users to create vector databases of local documents, and then perform Q&A through the existing DBQA function;
4. The new integrated interface not only directly supports group chat and single-model chat, but also allows you to import Prompt Sets or upload files at any time, and can also be used for related RAGs.

Welcome to experience it and give us feedback. If you encounter any difficulties, please feel free to contact us 😁

Kuwa v0.3.0-beta1 download info: https://kuwaai.org/blog/kuwa-os-v0.3.0-beta1
Kuwa official website: https://kuwaai.org/

New Contributors

• @cclljj made their first contribution in #10

Full Changelog: v0.2.1-taide...v0.3.0-beta1

---

社群的朋友大家好，

依規劃，我們推出 kuwa-v0.3.0-beta1 版本讓各位搶先體驗😁

改了什麼

這版主要新增了 Bot、Store、RAG toolchain等功能，以及新的聊天與群聊整合介面：

1. Bot 讓使用者可以no code建立Bot應用，可調整System prompt來實現如角色扮演、執行特定任務等不同功能，支援部份的Ollama modelfile設定；
2. Store 讓使用者可以自行建置及維護共用的Bot應用商店，使用者間也可以共享Bot；
3. RAG toolchain 讓使用者可以將本機文件建立成自有的向量資料庫，再透過即有的 DBQA 功能進行問答；
4. 新的整合介面不僅直接支援群聊與單模型聊天，隨時可以匯入Prompt Sets或上傳檔案，也可以用於相關RAG。

歡迎各位體驗之後給予回饋，遇到任何困難也歡迎與我們聯絡😁

Kuwa v0.3.0-beta1 下載資訊: https://kuwaai.org/zh-Hant/blog/kuwa-os-v0.3.0-beta1
Kuwa官網: https://kuwaai.org/zh-Hant/

新貢獻者

• @cclljj 做出了第一次貢獻 #10

完整變更紀錄: v0.2.1-taide...v0.3.0-beta1

Release Kuwa TAIDE v0.2.1

Hello everyone,

The TAIDE model released the Llama3-TAIDE-LX-8B-Chat-Alpha1 version today. Friends who use the Kuwa TAIDE version only need to update to the latest v0.2.1 version to experience the latest version of the TAIDE model. In addition to updating the TAIDE model, this version also expands the support for local models and fixes some minor problems, hoping to provide everyone with a better user experience.

The detailed update content is as follows:

1. Integrate GPU and CPU versions
2. Fix the line break bug when exporting chat records
3. Fix the text display error of the copy button
4. Fix the chat record disappearing bug caused by incorrect group chat sorting
5. HuggingFace executor adds support for various new on-premise models recently launched by vendors such as Apple and Microsoft

kuwa-taide-v0.2.1 single executable file download information: https://kuwaai.org/blog/kuwa-os-v0.2.1-taide
Manual model update tutorial: https://kuwaai.org/blog/migrate-to-kuwa-os-v0.2.1-taide
Llama3-TAIDE-LX-8B-Chat-Alpha1 release post: TAIDE Official Facebook Page Post
Kuwa official website: https://kuwaai.org/

---

大家好，

TAIDE 模型今天釋出了 Llama3-TAIDE-LX-8B-Chat-Alpha1 版，
使用 Kuwa TAIDE 版本的朋友只要更新到最新的 v0.2.1 版本就能體驗最新版的 TAIDE 模型，
這個版本除了更新 TAIDE 模型外，還擴充了地端模型支援，並修正一些小問題，希望能讓大家有更好的使用體驗。

詳細更新內容如下：

1. 將GPU、CPU版本整合
2. 修正匯出聊天紀錄的換行bug
3. 修正複製按鈕的文字顯示錯誤
4. 修正群聊排序錯誤導致的聊天紀錄消失bug
5. HuggingFace executor新增支援Apple、Microsoft等業者近期推出的多種地端模型

kuwa-taide-v0.2.1 單一執行檔下載資訊: https://kuwaai.org/zh-Hant/blog/kuwa-os-v0.2.1-taide
手動更新模型教學: https://kuwaai.org/zh-Hant/blog/migrate-to-kuwa-os-v0.2.1-taide
Llama3-TAIDE-LX-8B-Chat-Alpha1 釋出貼文: TAIDE 官方 Facebook 粉絲團貼文
Kuwa官網: https://kuwaai.org/zh-Hant/

Full Changelog: v0.2.0-taide...v0.2.1-taide

Release Kuwa TAIDE v0.2.0

Dear friends in the community,

The TAIDE model was released today, and we are happy to also release a customized Kuwa system for Windows with the built-in TAIDE LX 7B Chat 4bit model.

This customized system is a self-extracting single executable file, and the TAIDE model is built in as the default local model option, which can be executed in CPU and GPU environments, allowing everyone to quickly and easily experience the effects of the TAIDE model and perform related GenAI applications.

In addition, this system is customized based on the previously released v0.2.0-beta, so it can also use the TAIDE model for RAG applications. However, it should be noted that RAG will generate longer inputs, and it is recommended to use the GPU version for inference.

The Kuwa system and TAIDE model are still under continuous development and improvement, and there may be some instability. The content generated by this system is for reference only, and its correctness is not guaranteed. Users still need to verify it themselves. Please do not make inappropriate dialogue content public to avoid unexpected troubles.

kuwa-taide-v0.2.0 single executable file download information: https://kuwaai.org/zh-Hant/blog/kuwa-os-v0.2.0-taide
TAIDE official website: https://taide.tw/
Kuwa official website: https://kuwaai.org/zh-Hant/

---

社群的朋友大家好，

TAIDE 模型於今天釋出，我們很開心也能釋出內建 TAIDE LX 7B Chat 4bit 模型的 Windows 版客製化 Kuwa 系統。

此客製化系統為自解壓縮的單一執行檔，並將 TAIDE 模型內建成預設的地端模型選項，可在 CPU 與 GPU 環境下執行，讓大家可以方便快速的體驗 TAIDE 模型的效果，並進行GenAI的相關應用。

此外，此系統基於稍早釋出的 v0.2.0-beta 進行客製化，因此也可以使用 TAIDE 模型進行 RAG應用。惟須注意 RAG 會產生較長的輸入，建議使用 GPU 版本進行推論。

Kuwa 系統及 TAIDE 模型仍還在持續開發及改善，難免會發生不穩。本系統生成的內容僅供參考，不擔保其正確性，仍需使用者再行查證；請勿將不適的對話內容公開，以免帶來不預期的困擾。

kuwa-taide-v0.2.0 單一執行檔下載資訊: https://kuwaai.org/zh-Hant/blog/kuwa-os-v0.2.0-taide
TAIDE官網: https://taide.tw/
Kuwa官網: https://kuwaai.org/zh-Hant/

Release v0.2.0-beta

Release Note v0.2.0-beta

Release Note

Windows Portable Edition

1. Adjust the model hosting method:
   • Enable Gemini Pro and ChatGPT APIs by default
   • Use Gemini Pro by default to launch RAG applications for WebQA and DocQA
2. Deprecate env.bat and use run.bat to launch the executor instead
3. Enhance executor functionality:
   • Allow direct configuration of execution instructions, parameters, and other information
   • Adjust init.bat to be a simple tool to help create run.bat. Users can also directly write run.bat to launch the required model
4. Fix the error of a non-existent PHP download link in v0.1.0 (archived due to version update)
5. Integrate RAG into the simplified launch framework of the Windows version
6. Specify file path improvements:
   • In the executors folder of the Windows version, files will be specified using relative paths by default
7. Fix the executor error of the Custom category
8. Permission adjustment:
   • Only groups with Manage Tab permission can be directly assigned permission to use the model when the model is added
9. Fix the issue in the Windows version where Redis uses localhost as the IP, causing DNS queries to be delayed by 2 seconds each time

Docker Edition

1. Integrate RAG (DocQA / WebQA / DB QA) into the executor's Docker image and compose the system
2. Provide a compose example of Gemini Pro using a global API key
3. Complete the missing words in the document

Executor

1. Provide a TAIDE API executor, which can be directly connected to the TAIDE API of TECO
2. Port RAG executor (DocQA / WebQA / DB QA / SearchQA) to the new framework
3. Let the RAG executor support automatic model search, i18n, and interrupt generation

Kernel

1. Provide an API to list currently available executors

Multi-chat

1. Adjust the timeout waiting time for the no-message state:
   • Extend from 1 minute to 2 minutes to accommodate the waiting time for the RAG processing speed gap
2. Fix the bug of AdminSeeder:
   • Fix the issue of granting duplicate model usage permissions
3. Add the function of sending Kuwa tokens from the website to the executor:
   • To fix the past issue of unowned API tokens for RAG
4. Add a method to adjust the default model image path:
   • Can be configured via LLM_DEFAULT_IMG in the .env file
5. Fix the bug that the API in v0.1.0 could not be used normally

Known Issues and Limitations

1. Since RAG-related applications may generate longer inputs, it is easy to cause timeout errors if they are connected to CPU models on the local side. It is suggested to connect to cloud models or use GPU-based local models

v0.2.0-beta 發行說明

發行說明

Windows Portable 版

1. 調整模型架設方式：
   • 預設啟用 Gemini Pro 與 ChatGPT 的 API
   • 預設使用 Gemini Pro 來啟動 WebQA 與 DocQA 的 RAG 應用
2. 廢除 env.bat，改用 run.bat來啟動executor
3. 改進 executor 功能：
   • 允許直接設定執行指令、參數等資訊
   • 調整 init.bat 作為單純輔助建立 run.bat 的工具，使用者也可直接撰寫run.bat來啟動所需模型
4. 修正 v0.1.0版 PHP 下載連結不存在的錯誤（因版本更新而被封存）
5. 整合 RAG 至 Windows 版本簡易啟動框架下
6. 指定檔案路徑改善
   • 在 Windows 版本的 executors 資料夾下，檔案預設會使用相對路徑指定
7. 修正 Custom 類別的 executor 錯誤
8. 權限調整：
   • 只有管理 Tab 權限的群組可以在模型新增的時候直接被賦予該模型使用權限
9. 修正 Windows 版本中 Redis 因使用 localhost 作為 IP 導致 DNS 查詢每次需延遲 2 秒的問題

Docker 版

1. 整合 RAG (DocQA / WebQA / DB QA) 至 executor 的 Docker image 與 compose 系統中
2. 提供 Gemini Pro 使用全域 API key 的 compose 範例
3. 補完文件缺漏字句

Executor

1. 提供 TAIDE API executor，可直接串接國網的 TAIDE API
2. 移植 RAG executor (DocQA / WebQA / DB QA / SearchQA) 到新版框架上
3. 讓 RAG executor 支援自動模型尋找、i18n與中斷生成

Kernel

1. 提供列出目前可用 executor 的 API

Multi-chat

1. 調整無訊息狀態的 timeout 等待時間：
   • 從 1 分鐘延長至 2 分鐘，以因應 RAG 處理速度的空窗等待時間
2. 修復 AdminSeeder 的 bug：
   • 修正給予重複的模型使用權限的問題
3. 新增網站傳送 kuwa token 到 executor 上的功能：
   • 用以修正 RAG 過去無歸屬的 API token 問題
4. 增加調整預設模型圖片路徑的方法：
   • 可透過 .env 檔案的 LLM_DEFAULT_IMG 進行設定
5. 修正 v0.1.0 的 API 無法正常使用的 Bug

已知問題及限制

1. RAG 相關應用因會產生較長的輸入，若串接地端 CPU 模型容易產生超時錯誤，建議串接雲端模型，或是使用 GPU 的地端模型

Release v0.1.0

Release Notes v0.1.0

Windows Portable Version

1. Adjusted model setup for easier configuration of multiple models.
2. Fixed various errors occurring when using MySQL or PostgreSQL.
3. Readme updated for improved completeness.

Docker Version

1. Added ability to start the entire system and multiple Executors with Docker Compose.
2. Uses stable software stack suitable for production environments.
3. Modular design allows selection of Executor types and quantity to start.

Executor

1. Added command-line interface launcher for easy start-up of multiple Executors. Allows common parameters such as Prompt templates, System Prompt, and Generation config to be passed as commands.
2. Supports popular on-premise model inference frameworks like Huggingface Transformers and Llama.cpp.
3. Compatible with inference services such as vLLM, LiteLLM, that work with OpenAI API or Gemini-Pro API.
4. Packaged common features into Executor framework, such as automatic retry registration, automatic expiration, automatic history trimming, and interrupt generation.
5. Packaged Executor framework into a package for easy extension.
6. Fixed bug in llama.cpp executor's generation error.
7. Upgraded underlying framework to FastAPI for improved efficiency and stability.

Multi-chat

1. Fixed bug causing the website to jump to the /stream route.
2. Added default images for models.
3. Fixed some minor bugs.
4. Added more command-line tools for website configuration.

Full Changelog: https://github.com/kuwaai/genai-os/commits/v0.1.0

v0.1.0 發行說明

Windows Portable 版

1. 模型架設方式調整，現在可以簡易的設定更多模型
2. 修正使用 MySQL 或 PostgreSQL 時會發生的多種錯誤
3. Readme 更新的更完整

Docker 版

1. 可透過 Docker Compose 一鍵啟動整個系統與多個 Executor
2. 選用穩定的軟體堆疊，可直接作為生產環境
3. 模組化設計，可自由選擇要啟動的 Executor 種類與數量

Executor

1. 加上命令列介面啟動器，可一鍵啟動多種 Executor，並允許將Prompt模板、System Prompt、生成設定 (Generation config)等常用的參數用指令帶入
2. 支援Huggingface Transformers與Llama.cpp等常見地端模型推論框架
3. 支援與 OpenAI API 或 Gemini-Pro API 相容的推論服務，如vLLM、LiteLLM等
4. 將常用功能包成 Executor 框架，如自動註冊重試、自動註銷、自動歷史紀錄修剪、中斷生成等
5. 將 Executor 框架打包成套件，方便自行擴充 Executor
6. 修正 llama.cpp executor 的生成錯誤 bug
7. 底層框架換成 FastAPI 以提昇效率與穩定性

Multi-chat

1. 網站機率跳到/stream route 的 bug 修正
2. 模型預設圖片
3. 一些小 Bug 修復
4. 多了更多設定網站用的 Command line 工具

完整更新紀錄: https://github.com/kuwaai/genai-os/commits/v0.1.0