v0.3.2

What's Changed

1. support llama-2 model @fangyinc
2. add autoDL image and faq @Aries-ckt
3. add dbgpt api and you can use dbgpt according pip install dbgpt @Aries-ckt
4. structured prompt data interaction between dbgpt-server and llm-server @fangyinc

新特性

1. 支持llama-2系列模型
2. 提供AutoDL镜像
3. dbgpt api v0.3.0
4. 结构化prompt

v0.3.1

What's Changed

1. Solving the chatglm2-6b Multi GPUS Reasoning Problem @qutcat1997
2. chatglm not working in doc qa, meta qa and plugin @fangyinc

新特性

1. 解决chatglm2-6b多卡推理问题 @qutcat1997
2. chatglm 模型在知识问答, 数据库问答优化 @fangyinc

v0.3.0

What's Changed

1. brand new web UI for DB-GPT @changhuiping.chp @shiweisong.ssw
2. refactor dbgpt-server, provide open API for chat according to the scenario @yhjun1026 @Aries-ckt
3. prompt memory @yhjun1026
4. refactor knowledge embedding api, provide knowledge back-end api @Aries-ckt

新特性

1. 全新的web-ui，全新的体验@changhuiping.chp @shiweisong.ssw
2. dbgpt-server重构，按照对话场景提供chat的openapi @yhjun1026 @Aries-ckt
3. prompt对话记忆，案例加载@yhjun1026
4. 知识库api按照知识类型进行调整，提供知识库后端api @Aries-ckt

v0.2.3

What's Changed

1. Add LLM Support for ChatGLM2-6b、ChatGLM-6b-int4 @csunny

新增特性

1. 支持ChatGLM2-6b、ChatGLM2-6b-int4模型。 @csunny

v0.2.2

What's Changed

• DB-GPT v0.2.2 is released, mainly updating the following functions:
  a. Multi-model management supports gpt4all-6b, which can run on M1/M2 and CPU. #178 @sheri528
  b. ChatDB capability of vectors and DBs, conversations can be carried out based on the metadata information of the database. @Aries-ckt
  c. The plug-in model is built for ease of use, and the Chat2Dashboard capability is built in by default. @yhjun1026
  d. Added Weaviate support for unified vector storage. #222 @younis-ba
  e. Multi GPU support #203 #207 @zhanghy-sketchzh
  f. Add ChatGPT API Proxy #144 @xuyuan23

• DB-GPT v0.2.2 版本发布，主要更新以下功能：
  a. 多模型管理支持gpt4all-6b, 可以在M1/M2以及CPU上运行。
  b. 基于向量与DB的ChatDB能力发布，可以根据数据库元数据信息进行对话。
  c. 插件模型易用性建设，默认内置了Chat2Dashboard能力.
  d. 统一向量存储新增Weaviate支持。
  e. 多GPU推理支持。
  g. 支持OpenAI proxy代理

v0.2.1

What's Changed

• llm model manage support gorilla #173 @zhanghy-sketchzh
• llm model manage support falcon #175 @zhanghy-sketchzh
• documents chart embedding index error fix #77 @Aries-ckt
• position adjustment, LLM nativate chat by default @csunny

v0.2.0

What's Changed

• On the basis of the Vicuna-13B base model, task chain invocation is implemented based on plugins. For example: One-shot database creation. @yhjun1026 @xuyuan23 #125
• Add effect comparison evaluation, proxy ChatGPT. Implement evaluation of the effect of the private deployment model and the ChatGPT model. @csunny #144
• Database metadata asset embedding and prompt templates @Aries-ckt #102
• DB-GPT-ByteBase-Plugin being proposed to enable the use of third-party plugins and have the ability to use external tools @xuyuan23 @yhjun1026 #120
• Add the Guanaco family model to the large model management system, which supports running the Guanaco-33B on 24GB graphics memory @csunny @zhanghy-sketchzh #133

V0.1.1

What's Changed

• supports multiple language modes, and the language can be set in the configuration file .env.template (currently supports Chinese [zh] and English [en]) @xuyuan23

v0.1.0

What's Changed

• Built-in database query plugin, supports automatic data querying. @yhjun1026
• Support milvus vector storage. @Aries-ckt
• Update the docs use readthedocs. @csunny
• Add lint for source code @yihong0618

Docs add

Add docs for DB-GPT project. the project url is there: db-gpt.readthedocs.io