0.2.0b5

Highlights

🔥 Experimental GPTAssistantAgent, which leverages the OpenAI Assistant API for conversational capabilities in AutoGen. This agent is unique in its reliance on the OpenAI Assistant API for state management, differing from other agents using the Completion API.

• Blogpost: https://microsoft.github.io/autogen/blog/2023/11/13/OAI-assistants/
• Notebook 1: GPTAssistantAgent in a hello-world example: https://github.com/microsoft/autogen/blob/main/notebook/agentchat_oai_assistant_twoagents_basic.ipynb
• Notebook 2: GPTAssistantAgent using function call: https://github.com/microsoft/autogen/blob/main/notebook/agentchat_oai_assistant_function_call.ipynb
• Notebook 3: GPTAssistantAgent with code interpreter: https://github.com/microsoft/autogen/blob/main/notebook/agentchat_oai_code_interpreter.ipynb
• Notebook 4: GPTAssistantAgent in a group chat: https://github.com/microsoft/autogen/blob/main/notebook/agentchat_oai_assistant_groupchat.ipynb

🔥 Blogpost for EcoAssistant, which is designed to solve user queries more accurately and affordably using assistant hierarchy and solution demonstration.

Thanks to @IANTHEREAL @gagb @LeoLjl @kevin666aa @qingyun-wu @sonichi @JieyuZ2 and all the other contributors!

What's Changed

• Fix docstring of get_or_create by @thinkall in #583
• Refactor GPTAssistantAgent by @gagb in #632
• uncomment test_hierarchy_flow_using_select_speaker by @qingyun-wu in #640
• Add basic notebook for gptassistant by @gagb in #636
• Openai assistant function usage notebook by @IANTHEREAL in #639
• bump version by @sonichi in #645
• add EcoAssistant blog by @JieyuZ2 in #647
• Add code interpreter for GptAssistant Notebook. by @LeoLjl in #644
• Add openai_client as property and update notebook. by @LeoLjl in #654
• fix tag in EcoAssistant blog by @JieyuZ2 in #657
• Add agentchat_oai_assistant_groupchat.ipynb by @LeoLjl in #656
• Add blog post for open ai assistants by @gagb in #638
• add doc about openai assistants by @qingyun-wu in #659

New Contributors

• @LeoLjl made their first contribution in #644

Full Changelog: v0.2.0b4...0.2.0b5

v0.2.0b4

Highlights

• CompressibleAgent (experimental) can be used to handle long conversations. Notebook: https://github.com/microsoft/autogen/blob/main/notebook/agentchat_compression.ipynb
• Introducing Experimental GPT Assistant Agent: https://github.com/microsoft/autogen/blob/main/autogen/agentchat/contrib/gpt_assistant_agent.py. More improvements are under way.
• Experimental streaming support is added.
• seed is renamed to cache_seed to be compatible with openai's seed parameter in chat completion. Migration guide is updated: https://microsoft.github.io/autogen/docs/Installation/#migration-guide-to-v02
• Better warnings and error handling for group chat.

Thanks to @IANTHEREAL @kevin666aa @Alvaromah and all the other contributors!

What's Changed

• seed -> cache_seed by @sonichi in #600
• Added link to the new notebook by @joshkyh in #594
• update return type of WolframAlphaAPIWrapper.run() by @1073710317 in #523
• news update by @sonichi in #609
• Add EcoAssistant to the research page by @JieyuZ2 in #612
• Add CompressibleAgent by @kevin666aa in #443
• add AutoGen paper info at the beginning of readme by @qingyun-wu in #621
• Update oai_completion.ipynb by @AndreasVolkmann in #623
• Added warnings for some GroupChat misconfigurations and selection errors by @afourney in #603
• Introducing Experimental GPT Assistant Agent in AutoGen by @IANTHEREAL in #616
• added twitter(X) banner + link to readme by @malikmmoaz in #615
• Enable streaming support for openai v1 by @Alvaromah in #597
• improve readme by @qingyun-wu in #630
• Handled possible unclear IndexError in ConversableAgent.last_message method by @jeezrick in #622
• Fix test error of compressible agent by @kevin666aa in #631

New Contributors

• @1073710317 made their first contribution in #523
• @JieyuZ2 made their first contribution in #612
• @AndreasVolkmann made their first contribution in #623
• @IANTHEREAL made their first contribution in #616
• @malikmmoaz made their first contribution in #615
• @Alvaromah made their first contribution in #597
• @jeezrick made their first contribution in #622

Full Changelog: v0.2.0b3...v0.2.0b4

v0.2.0b3

A quick release to switch to openai-python v1.1.1.
Thanks to @joshkyh for a new group chat notebook example with hierarchical flow: https://github.com/microsoft/autogen/blob/main/notebook/agentchat_hierarchy_flow_using_select_speaker.ipynb

What's Changed

• update version of openai dependency by @sonichi in #588
• Notebook/hierarchy flow by @joshkyh in #482

New Contributors

• @joshkyh made their first contribution in #482

Full Changelog: v0.2.0b2...v0.2.0b3

v0.2.0b2

This is a beta release of v0.2.0b2.

Highlights

• Support for GPT-4V! Introduced Large Multimodal Models in AgentChat, enhancing capabilities and interactions within the platform. Blogpost: https://microsoft.github.io/autogen/blog/2023/11/06/LMM-Agent
• Improved codebase reliability with updates such as dict copying before modifications (fixing a bug for Azure OpenAI) and various typo fixes.
• Added support for unstructured data in retrieve chat (RAG).
• Expanded functionality with async support for better get_human_input handling.
• A new simple Testbed tool for Autogen processes.
• Enhanced developer tools and documentation, including new README and TRANSPARENCY_FAQS updates.

Thanks to all the testers for the v0.2 migration. Thanks to @BeibinLi @sonichi @AkariLan @vatsalya-vyas @gfggithubleet @gagb @thinkall @hung-ngm @afourney @AaadityaG @jasondotparse @bonadio @aayushchhabra1999 @qingyun-wu @eltociear @marcgreen and other contributors!

What's Changed

• Added a simple Testbed tool for repeatedly running templated Autogen scenarios with tightly-controlled initial conditions. by @afourney in #455
• Fix typo import autogen by @hung-ngm in #549
• Add support to unstructrued by @thinkall in #501
• Update TRANSPARENCY_FAQS.md by @gfggithubleet in #492
• Update README.md by @vatsalya-vyas in #507
• fix wrong 'Langchain Provided Tools as Functions' doc ref by @AkariLan in #495
• copy dicts before modifying by @sonichi in #551
• Large Multimodal Models in AgentChat by @BeibinLi in #554

New Contributors

• @hung-ngm made their first contribution in #549
• @gfggithubleet made their first contribution in #492
• @vatsalya-vyas made their first contribution in #507
• @AkariLan made their first contribution in #495

Full Changelog: v0.2.0b1...v0.2.0b2

v0.2.0b1

This is a beta release of v0.2.0.

Highlights

• Switching to openai v1. Please read the migration guide and report bugs.
• Support async function execution & get_human_input.
• Improvements in documentation and notebooks.

Thanks to all the reviewers for the v0.2 migration. Thanks to @aayushchhabra1999 @bonadio @marcgreen and other contributors!
 

What's Changed

• Update Installation.md by @gagb in #456
• Update FAQ with workaround for Issue #251 by @marcgreen in #405
• Fix typo in README.md by @eltociear in #481
• Fix/async function and tool execution by @aayushchhabra1999 in #87
• Adding async support to get_human_input by @bonadio in #466
• Added example .txt file for agentchat_langchain sample notebook by @jasondotparse in #373
• Fix : Typo by @AaadityaG in #506
• Dev/v0.2 by @sonichi in #393

New Contributors

• @marcgreen made their first contribution in #405
• @aayushchhabra1999 made their first contribution in #87
• @bonadio made their first contribution in #466
• @jasondotparse made their first contribution in #373
• @AaadityaG made their first contribution in #506

Full Changelog: v0.1.14...v0.2.0b1

v0.1.14

Highlights

• 👀Give vision to your agent: multimodal examples are added at https://github.com/microsoft/autogen/blob/main/notebook/agentchat_lmm_llava.ipynb.
• 📖TeachableAgent blogpost: https://microsoft.github.io/autogen/blog/2023/10/26/TeachableAgent.
• 🧵 Run a chat in a different thread/process: using thread safe timeout for code execution.
• 🪶 Qdrant vector store: A QdrantRetrieveUserProxyAgent is added in contrib/.
• 🚦 Support new version of chromadb in retrieve chat.
• 🧮 Token count utils.
• 🐋 Improve vscode extension setup in codespace.
• 📝 Many improvements in documentation, FAQ, useful tips, such as
  • how to prevent gpt-3.5 agents' appreciation loop
  • fixes in the langchain notebook
  • link to the roadmap
  • common issues in retrieve chat

Thanks to @BeibinLi @rickyloynd-microsoft @ragyabraham @Anush008 @thinkall @kevin666aa @shruti222patel @craigomatic @AaronWard and all the other contributors!

Headsup

v0.2 release is near the corner, and we'll switch to openai v1 in it. Please check for breaking changes in #393. We'll try to add as many features back as possible before the release. If you see any breaking changes that affect your work, please comment in the PR thread.

What's Changed

• Added Roadmap to Getting-Started.md by @Gourav2609 in #324
• Resolves Typo Correction #336 by @2good4hisowngood in #338
• Update README.md by @IshitaPathak in #321
• Updating Examples to follow new categorical structure. #273 by @rabbabansh in #327
• Fix grammar in agentchat_auto_feedback_from_code_execution.ipynb by @shruti222patel in #342
• Fix spellings in notebooks by @shruti222patel in #341
• Fix grammar and spelling in agentchat_chess.ipynb by @shruti222patel in #346
• Improve readability for agentchat_function_call.ipynb by @shruti222patel in #347
• Improve readability in agentchat_groupchat_research.ipynb by @shruti222patel in #350
• fix format issue by @qingyun-wu in #358
• Improve readability in oai_openai_utils.ipynb by @shruti222patel in #365
• Improve readability in oai_completion.ipynb by @shruti222patel in #364
• Fix grammar in agentchat_web_info.ipynb by @shruti222patel in #362
• Improve readability for agentchat_two_users.ipynb by @shruti222patel in #361
• Fix the grammar and spelling in agentchat_human_feedback.ipynb by @shruti222patel in #354
• Fix spelling and grammar in agentchat_langchain.ipynb by @shruti222patel in #355
• Fix grammar and spelling in agentchat_planning.ipynb by @shruti222patel in #356
• Fix grammar in agentchat_stream.ipynb by @shruti222patel in #357
• headsup about dependency change by @sonichi in #378
• Add link to OptiGuide #371 by @smty2018 in #376
• Update docs in RetrieveChat notebook and Installation by @thinkall in #379
• config list for test by @sonichi in #395
• Supporting MultiModal Models: an example with LLaVA Notebook by @BeibinLi in #286
• using thread safe timeout to allow code execution to be compatible wi… by @ragyabraham in #224
• Fix tmp dir not exists by @thinkall in #401
• feat: Qdrant vector store support by @Anush008 in #303
• [Blocking Issue] Add tests dependencies for qdrant and fix chromadb errors by @thinkall in #435
• TeachableAgent blog post by @rickyloynd-microsoft in #436
• Adds jupyter as a vscode extension, fix validation errors in devcontainer.json by @craigomatic in #433
• Update FAQ section in documentation by @AaronWard in #390
• Add token_count_util by @kevin666aa in #421
• spelling fix for math_user_proxy_agent.py by @codingdrone in #431
• bump version to 0.1.14 by @sonichi in #400

New Contributors

• @2good4hisowngood made their first contribution in #338
• @IshitaPathak made their first contribution in #321
• @rabbabansh made their first contribution in #327
• @ragyabraham made their first contribution in #224
• @Anush008 made their first contribution in #303
• @craigomatic made their first contribution in #433
• @codingdrone made their first contribution in #431

Full Changelog: v0.1.13...v0.1.14

v0.1.13

Highlights

A preliminary TeachableAgent is added to allow users to teach their assistant facts, preferences, and tasks unrelated to code generation. Example notebook: https://github.com/microsoft/autogen/blob/main/notebook/agentchat_teachability.ipynb

Conversational assistants based on LLMs can remember the current chat with the user, and can even demonstrate in-context learning of things that the user teaches the assistant during the chat. But these memories and learnings are lost once the chat is over, or when a single chat grows too long. In subsequent chats, the user is forced to repeat any necessary instructions over and over.

TeachableAgent addresses these limitations by persisting user teachings across chat boundaries in long-term memory (a vector database). Memory is saved to disk at the end of each chat, then loaded from disk at the start of the next. Instead of copying all of memory into the context window, which would eat up valuable space, individual memories (called memos) are retrieved into context as needed. This allows the user to teach frequently used facts, preferences and skills to the agent just once, and have the agent remember them in later chats.

This release also contains an update about openai models and pricing, and restricts the openai package dependency version. In v0.2 we will switch to openai>=1.

Thanks to @rickyloynd-microsoft @kevin666aa and all the other contributors!

What's Changed

• doc: Update FAQ.md by @haseeb-xd in #282
• Fix format and links in documentations by @thinkall in #310
• update openai models and pricing by @kevin666aa in #316
• guidance for contribution by @sonichi in #320
• TeachableAgent by @rickyloynd-microsoft in #278
• add downloads stats in readme by @sonichi in #334
• bump version to 0.1.13 by @sonichi in #333

New Contributors

• @haseeb-xd made their first contribution in #282
• @rickyloynd-microsoft made their first contribution in #278

Full Changelog: v0.1.12...v0.1.13

v0.1.12

This release contains a significant improvement to function call in group chat. It decreases the chance of failures for group chat involving function calls. It also contains improvements to RAG agents, including added support for custom text splitter, example notebook for RAG agent in group chat, and a blogpost. Thanks to @thinkall and other contributors!

What's Changed

• Add support to custom text spliter by @thinkall in #270
• Update FAQ.md, elaborate on how to customise docker image and pick 'python:3' to solve typical errors by @maxim-saplin in #269
• Add group chat and retrieve agent example by @thinkall in #227
• docs: added virtual environment setup process by @Ayushpanditmoto in #249
• Fixing code execution broken link at Getting Started page by @maxim-saplin in #275
• Bump @babel/traverse from 7.20.1 to 7.23.2 in /website by @dependabot in #283
• Add a blog for RAG agents by @thinkall in #284
• function call filter in group chat by @sonichi in #294
• fix typo in website/blog/2023-05-18-GPT-adaptive-humaneval/index.mdx by @suravkshrestha in #299
• Organize notebooks into logical groups #273 by @victordibia in #288
• Update docs for extra dependencies and virtual env creation by @thinkall in #297
• Update README.md with Roadmap. by @gagb in #304
• bump version to 0.1.12 by @sonichi in #301

New Contributors

• @maxim-saplin made their first contribution in #269
• @Ayushpanditmoto made their first contribution in #249
• @suravkshrestha made their first contribution in #299
• @victordibia made their first contribution in #288

Full Changelog: v0.1.11...v0.1.12

v0.1.11

This release contains bug fixes, more clear behaviors for docker, and model compatibility improvement. Agents using Langchain tools are demonstrated in https://github.com/microsoft/autogen/blob/main/notebook/agentchat_langchain.ipynb. Thanks to @ElliotWood and all the other contributors!

What's Changed

• Fix typo in agentchat_MathChat.ipynb by @eltociear in #191
• Make getting started a little easier by @wayliums in #180
• Fix edge cases extracting code when the message's content is None by @msaelices in #205
• Add md for faqs by @gagb in #194
• Fix typo in README.md by @creator0131 in #193
• Icons not centered in the "autogen_agentchat.png" by @Steellgold in #169
• Addresses issue #199 by @afourney in #200
• Improving logging in oai.completion to show token_count by @kevin666aa in #179
• Update Getting-Started.md by @javidjamae in #213
• Added comment about config_list in the README example #216 by @smty2018 in #218
• fix: be compatible with customizing model without all fields in the response by @jtsang4 in #222
• Update Installation.md-with detailed explanation on add a period or newline #219 by @05hiteshbansal in #231
• Spelling fixes. by @shruti222patel in #220
• Typo by @qingyun-wu in #234
• bump version to 0.1.11 by @sonichi in #242
• docstr updated for use_docker in execute_code by @Gourav2609 in #233
• document retrieve chat by @olaoluwasalami in #258
• Add a figure to autogen notebook #215 by @smty2018 in #246
• Langchain tool bridge (#262) by @sonichi in #263

New Contributors

• @wayliums made their first contribution in #180
• @msaelices made their first contribution in #205
• @gagb made their first contribution in #194
• @creator0131 made their first contribution in #193
• @Steellgold made their first contribution in #169
• @javidjamae made their first contribution in #213
• @smty2018 made their first contribution in #218
• @jtsang4 made their first contribution in #222
• @05hiteshbansal made their first contribution in #231
• @shruti222patel made their first contribution in #220
• @Gourav2609 made their first contribution in #233

Full Changelog: v0.1.10...v0.1.11

v0.1.10

This release adds support to plug in customized vectordb and embedding functions.

What's Changed

• Bump version to 0.1.10 by @thinkall in #181
• Add support to customized vectordb and embedding functions by @thinkall in #161

Full Changelog: v0.1.9...v0.1.10