v0.1.1

What's Changed

• Fix the bug that the max_retries doesn't work in ReActAgent
• Fix the bug that setting mutliple configurations with different dahscope api keys will result in only one api key taking effect
• Support nested Pydantic BaseModel objects within the parser
• Support token couting in AgentScope for OpenAI (both vision and non-vision llms), Gemini, DashScope APIs, and support to register token counting functions locally
• Fix bug in streaming mode
• Adjust the basic prompt engineering strategy within the format function to avoid misleading
• [Experimental]Add environment modules

Details

• Update paper list by @pan-x-c in #440
• [Hot Fix] Add test for minimal dependencies by @pan-x-c in #445
• Change the role to "user" in format function of ollama chat model by @zhijianma in #443
• Remove something by @rayrayraykk in #454
• Fix ut by @rayrayraykk in #456
• Slove the format_map error in router_agent.py by @yizhangliu in #436
• docs: add README_JA.md by @eltociear in #455
• [Hot Fix] Fix stream text by @pan-x-c in #461
• Make CACHE_DIR of agentscope studio configurable by @rayrayraykk in #467
• [Hotfix] minor fix on embedding related bugs by @ZiTao-Li in #463
• [HOTFIX] Fix the error in format function by adding system message by @DavdGao in #472
• Minor fix on max_iters parameter passing bug by @zzzmc in #464
• use model_dump instead of dict(*) in MarkdownJsonDictParse by @wujiren in #471
• Add environment modules by @pan-x-c in #378
• Fix typos in environment modules by @pan-x-c in #474
• Support token counting for OpenAI, DashScope, Gemini and HuggingFace models by @DavdGao in #470
• update dashscope version requirement by @cmgzn in #477
• [HOTFIX] Fix the error when using multiple dashscope api keys by @DavdGao in #478
• [Hotfix] change postAPI embedding model _parse_resonse to OpenAI format by @ZiTao-Li in #479

New Contributors

• @zzzmc made their first contribution in #464
• @wujiren made their first contribution in #471
• @cmgzn made their first contribution in #477

Full Changelog: v0.1.0...v0.1.1

v0.1.0

What's Changed

Model APIs

1. Support ZhipuAI API (#181)
2. Support Yi API (#343)
3. Support streaming mode in AgentScope (#347)
4. Support LiteLLM (#204)
5. Format input messages according to the name of the model (#381)

Service

Add the following new service functions

1. TripAdvisor search
2. Dblp search
3. wikipedia search
4. Web Browser Control
5. OpenAI and DashScope multimodal service functions

Others

1. Implement MoA method in AgentScope (#327)
2. A new prompt engineering module (#242)
3. Improve RAG module (#184)

Details

• Add ZhipuAI model wrappers by @garyzhang99 in #181
• 【HOTFIX】Fix error in gomoku example by @DavdGao in #195
• [HOTFIX] Fix typo in 0.0.4 README_ZH by @pan-x-c in #199
• Implement DBLP Search Functionality by @PengHongyiNTU in #177
• Add banner by @rayrayraykk in #201
• [DOC] Update distributed_basic/README.md by @eltociear in #207
• [HOTFIX] Fix bug in board agent by @DavdGao in #212
• [Hotfix] Fix memory loading error by @ZiTao-Li in #197
• [HOTFIX] Update prompt strategy in Ollama chat API. by @DavdGao in #206
• Add litellm wrapper by @garyzhang99 in #204
• Added README file for conversation_self_organizing by @zyzhang1130 in #210
• update issues temp by @rayrayraykk in #216
• Add support for vision models in OpenAI Chat API, e.g. gpt-4o, gpt-4-turbo, etc. by @DavdGao in #221
• [Hot Fix] Add ImportErrorReporter to optimize the error message when packages are missing by @pan-x-c in #224
• Add survey by @rayrayraykk in #223
• Example for SWE agent by @garyzhang99 in #173
• support dalle on workstation by @rayrayraykk in #225
• Make the dictionary parser much more flexible, and decouple task-specific information from DictDialogAgent by @DavdGao in #203
• [HOTFIX] Fix format error in tutorial by @DavdGao in #226
• Add an example on how to wrap langchain tools, dashscope api calls to tools that can be accessed in toolkit. by @PengHongyiNTU in #219
• [HOTFIX] Fix bug in the format method of ollama and litellm by @DavdGao in #230
• Doc/litellm doc by @garyzhang99 in #232
• Add as_server command and add server_id by @pan-x-c in #211
• 删除markdown “-”.md by @ZhuangXialie in #236
• Fix Precommit flake8/black conflict by @garyzhang99 in #241
• Add post API for embedding model and fix embedding model wrappers by @ZiTao-Li in #186
• add news for website maintenance by @rayrayraykk in #250
• [Hotfix] add version specification for the RAG example by @ZiTao-Li in #248
• Fix typo and prompt for game_werewolf by @qbc2016 in #251
• Gemini chat wrapper fix by @zyzhang1130 in #233
• [Hot Fix] Fix import error when import agentscope by @pan-x-c in #253
• Game gomoku tested model update by @zyzhang1130 in #247
• the cleaned-up version of game_werewolf_README branch by @zyzhang1130 in #246
• Add a demo video in README.md of both the repo and werewolf game by @DavdGao in #257
• [Hot Fix] Fix the bug that the save_dir is not effective by @pan-x-c in #263
• [HOTFIX] Fix the bug in loguru that cannot display special tags by @DavdGao in #266
• refine prompt for werewolf games by @qbc2016 in #267
• Update tested model list in SWE agent example by @zyzhang1130 in #271
• Ensure the game proceeds smoothly when some players declare abstentio… by @AskyJx in #272
• 修复了README文件的链接 by @KPCOFGS in #217
• Add DashScope API based multimodal service functions in AgentScope library by @PengHongyiNTU in #255
• [Hotfix] Add bs4 in setup.py to digest searched web pages by @qbc2016 in #274
• [Hotfix] Fix the error in unittest. by @DavdGao in #275
• Support Pydantic based validation within the MarkdownJsonDictParser by @DavdGao in #261
• Reformatted the README file of the example distributed_basic by @zyzhang1130 in #277
• Rebuild AgentScope Studio by @pan-x-c in #238
• Update tutorial for the new version by @DavdGao in #278
• Release v0.0.5 by @pan-x-c in #279
• [HOTFIX] Fix the import error in PyPI by @DavdGao in #280
• [HOTFIX] Fix error in setup.py and update doc by @DavdGao in #283
• Reformat and improve RAG module and agents by @ZiTao-Li in #184
• Pre-release 0.0.6a2 by @pan-x-c in #290
• [Hotfix] Add unit test and fix import error of LlamaIndexKnowledge by @ZiTao-Li in #289
• UT support more py ver by @rayrayraykk in #291
• fix(typo): update 208-distribute.md by @jianzs in #293
• Update files related to game werewolf by @qbc2016 in #285
• [Hotfix] Fix link issue in tutorial by @qbc2016 in #294
• Add prompt tuning modules by @garyzhang99 in #242
• [HOTFIX] Fix error in tutorial related to system prompt optimization by @DavdGao in #298
• [HOTFIX] Fix the error that cannot find rag in tutorial by @DavdGao in #299
• pre-release 0.0.6a3 by @pan-x-c in #300
• Change studio to gradio by @qbc2016 in #295
• update news: Workstation back online by @rayrayraykk in #305
• Add step mode for each examples in workstation by @qbc2016 in #284
• Update README.md for distributed_debate by @zyzhang1130 in #306
• Change domain for workstation by @qbc2016 in #316
• [Hotfix] Fix gradio displaying user message twice by @ZiTao-Li in #317
• Unify the typing of the agents' reply function and update their docstrings. by @DavdGao in #314
• Adding an example of setting up local embedding model by @ZiTao-Li in #322
• Fix paper ref names by @garyzhang99 in #323
• Remove Tht class from AgentScope and Unified typing into Msg rather than MessageBase by @DavdGao in #313
• Add OpenAI as services by @PengHongyiNTU in #304
• Fix display bug in workstation for group nodes by @PatrickG1014 in #325
• Fix env typo by @rayrayraykk in #329
• [HOTFIX] Remove notebook directory and fix typos. by @DavdGao in #330
• fix: 修改DashScopeWrapperBase在初始化模型配置时将空api_key赋值给dashscope.api_key by @xuanmiss in #335
• Initial commit for research on LLM-based algorithms by @yanxi-chen in #331
• [HOTFIX] Fix bug in DictDialogAgent by @DavdGao in #342
• Upload code for the research project on LLM-based algorithms by @yanxi-chen in #336
• If it is json(str) convert to json(dict) by @samchaolau in https://g...

v0.0.4

Features

• Agentscope Workstation and Copilot are online now
• More services
  • arxiv search
  • download from url
  • shell execution
• More models
  • dashscope multimodal
  • Llama3
• More examples
  • Conversation nl2sql
  • Llama3
  • Distributed Large Scale Simulation
• Refactor Agentscope Document

What's Changed

• Fix Dashscope total_token calculation by @pan-x-c in #146
• Add arxiv_search service function by @DavdGao in #148
• Fix "File exists error" when multiple agents are called at the same time by @pan-x-c in #147
• add dashscope multimodal by @qbc2016 in #92
• [HOTFIX] Support gemini.configure setting in model configuration by @DavdGao in #152
• Example for nl2sql by @garyzhang99 in #132
• [HOTFIX]Fix bug when dialog agent self.memory is None by @garyzhang99 in #157
• Docs/typo by @zhijianma in #155
• Add download_from_url service function by @DavdGao in #140
• Refine files in scripts. by @qbc2016 in #159
• Add drag-and-drop interface & add implementation of DAG running engine for drag-and-drop module by @rayrayraykk in #114
• Optimize the structure of Sphinx API doc by @pan-x-c in #162
• [HOTFIX] Fix bugs when x is None by @qbc2016 in #163
• Add llama3 support in example level by @DavdGao in #167
• [HOTFIX] Raise error when user's request is rejected by OpenAI safety system by @pan-x-c in #171
• Add checking reset before user input in studio by @qbc2016 in #154
• Add more service functions by @garyzhang99 in #156
• fix macos version error by @rayrayraykk in #174
• Add Compiler & Refactor workflow dag runner code by @rayrayraykk in #164
• Make workflow compiler return code string by @rayrayraykk in #178
• Add displaying cycle dots before agent reply in AS studio by @qbc2016 in #179
• [DOC] Generate .rst for each .py file and add init method into docs by @pan-x-c in #180
• Improve ReActAgent and extract several modules from it by @DavdGao in #153
• Update README with workstation info by @ZiTao-Li in #182
• Modify flask services by @qbc2016 in #172
• [HOTFIX]Fix the typos in parser module by @DavdGao in #189
• Agent servers support running any type of agent by @pan-x-c in #135

New Contributors

• @zhijianma made their first contribution in #155

Full Changelog: v0.0.3...v0.0.4

v0.0.3

What's Changed

• Fix the bug that the format method of ModelWrapperBase sub-classes cannot be used in distributed mode.
• Support running multiple Agent instances in a single Agent Server
• Optimize performance in distributed mode
• Update the ReActAgent

Details

• docs: fix error typo by @Undertone0809 in #129
• Update version number. by @DavdGao in #127
• Add agent_id for each agent instance by @pan-x-c in #128
• Remove WeChat QR code by @DavdGao in #134
• Update ReActAgent for qwen and other models by @DavdGao in #136
• Support agent_id in agent server by @pan-x-c in #94

New Contributors

• @Undertone0809 made their first contribution in #129

Full Changelog: v0.0.2...v0.0.3

v0.0.2

What's Changed

• Add ReActAgent agent
• Add web_digest service
• Add more examples
  • Autonomous Conversation with Mentions
  • Self-Organizing Conversation
  • Basic Conversation with LangChain library
  • Conversation with ReAct Agent
  • Conversation with RAG Agent
  • Gomoku
  • Distributed Debate
  • Distributed Parallel Search
• Add more model wrappers
  • DashScope: Chat, Image Synthesis, Text Embedding
  • Gemini: Chat, Embedding
  • Ollama: Chat, Embedding, Generation
• Add format function in model wrappers (chat and generation) to replace PromptEngine
• Add AgentScope Studio
• Update Documents accordingly

Details

• Add news in README.md by @DavdGao in #41
• Update README.md by @eltociear in #43
• Add Chinese README.md by @DavdGao in #42
• add tongyi model wrapper by @qbc2016 in #46
• add agent builder example by @Osier-Yi in #25
• Add win & macos support by @rayrayraykk in #48
• support for modelscope gradio by @qbc2016 in #44
• Stop using thread.name as uid, use a thread local value as uid by @qbc2016 in #51
• Fix bug in service_factory by @DavdGao in #52
• Optimize launcher of RPCAgent by @pan-x-c in #47
• [Ready] Translate tutorial into Chinese by @DavdGao in #49
• [Hot Fix]Fix text display in logger by @pan-x-c in #66
• Add DashScope model wrapper into AgentScope by @qbc2016 in #54
• Enhanced distributed debate example to support multiple users by @pan-x-c in #68
• Add ollama based local model services in AgentScope. by @DavdGao in #60
• Add groupchat with @ example by @rayrayraykk in #55
• [Hotfix] remove checking response.status_code by @qbc2016 in #72
• Correcting typos in documents by @wangwangbobo in #70
• add preprocess for roles for qwen by @qbc2016 in #73
• Add Gemini Chat and Embedding models in AgentScope by @DavdGao in #59
• [Hot Fix]Optimize error messages of ModelWrapper by @pan-x-c in #76
• Fix error in agent when use_memory is False by @DavdGao in #74
• Fix export issue: Add missing function ServiceFactory export by @zwjaaa in #79
• Add an example for the compatibility with langchain by @DavdGao in #62
• add hint for studio users by @rayrayraykk in #83
• Add unit testing for dashscope model by @qbc2016 in #85
• Update README.md by @ZiTao-Li in #75
• Hot fix in README.md by @DavdGao in #86
• Fix bug in basic conversation example by @DavdGao in #87
• Unify model implementation and clean up their code in AgentScope. by @DavdGao in #82
• [Hotfix] Fix hyperlinks by @qbc2016 in #89
• add AS studio usage in README by @qbc2016 in #81
• Fix the text_to_image_agent failure when accepting None input by @pan-x-c in #93
• [HotFix] Avoid unnecessary logging in studio by @DavdGao in #96
• [HotFix] Fix bug in dict_dialog_agent by @DavdGao in #98
• Fix bug in PromptEngine and provide prompt construction strategy for most chat and generation APIs by @DavdGao in #95
• Add AS studio in README by @qbc2016 in #90
• [HOTFIX] Create a new function to convert content to string. by @DavdGao in #100
• [Hotfix] Fix bug in format function by @DavdGao in #105
• Adding web page digest function to service module by @ZiTao-Li in #84
• [Hotfix] Fix bugs in AS studio by @qbc2016 in #106
• Adding gomoku example by @ZiTao-Li in #103
• Add attr participants to pipeline by @rayrayraykk in #107
• Fix Docs for v0.0.2 release by @pan-x-c in #108
• 📃 docs: 文档词语错误：永达->用法 by @xiaojinwhu in #110
• Fix typos in tutorial and add link to example model configurations by @DavdGao in #111
• Add a ReAct agent and an example conversation with this agent. by @DavdGao in #104
• fix mypy ignore pattern by @rayrayraykk in #115
• Fix bug in requiring necessary packages by @DavdGao in #123
• Introducing RAG module into package by @ZiTao-Li in #64
• Add distributed search example by @pan-x-c in #121
• Update sphinx doc for v0.0.2 by @pan-x-c in #125

New Contributors

• @eltociear made their first contribution in #43
• @Osier-Yi made their first contribution in #25
• @wangwangbobo made their first contribution in #70
• @zwjaaa made their first contribution in #79
• @xiaojinwhu made their first contribution in #110

Full Changelog: v0.0.1...v0.0.2

v0.0.1

What's Changed

• Add templates for issues and pull requests by @DavdGao in #1
• [DOC] add API reference documentation by @pan-x-c in #7
• add jupyter for conversation by @qbc2016 in #6
• Move defaults and types to constants by @pan-x-c in #9
• add distributed examples in notebook by @qbc2016 in #10
• add docs for all tutorial by @yxdyc in #12
• refactor monitor by @pan-x-c in #13
• [DOC] Update monitor tutorial and readme by @pan-x-c in #16
• change direct address to relative address for hyperlink in notebook by @qbc2016 in #15
• minor improvement for the links and position of "Tutorial"'s first appearance by @yxdyc in #17
• change hyperlink to ms by @qbc2016 in #20
• modify werewolf game by @qbc2016 in #11
• Fix tutorial/API_doc links in readme by @pan-x-c in #22
• Refactor Distributed Modules by @pan-x-c in #18
• Saving runtime configuration in root directory automatically by @DavdGao in #23
• refine prompt for werewolf game by @qbc2016 in #21
• [WEBUI] Add speak function within AgentBase. by @DavdGao in #27
• Refactor ModelWrapper by @pan-x-c in #24
• Refactor web ui interface. by @DavdGao in #29
• Add arXiv paper into README.md by @DavdGao in #33
• fix minor example bugs and typos by @garyzhang99 in #32
• Slight modification on README.md by @ZiTao-Li in #34
• Update description in setup.py by @DavdGao in #35
• Add ServiceFactory to support quick setup for tools model&agent. by @DavdGao in #26
• Fix module resource not available in win by @rayrayraykk in #38
• [Release] Preparation for the first release by @pan-x-c in #39

New Contributors

• @DavdGao made their first contribution in #1
• @pan-x-c made their first contribution in #7
• @qbc2016 made their first contribution in #6
• @yxdyc made their first contribution in #12
• @garyzhang99 made their first contribution in #32
• @ZiTao-Li made their first contribution in #34

Full Changelog: https://github.com/modelscope/agentscope/commits/v0.0.1